MASTER'S THESIS

High-temperature Compression Strength

of High-alumina Refractory Bricks Used in
Rotary Kilns of LKAB

Dmitrij Ramanenka

Master of Science in Engineering Technology

Materials Technology (EEIGM)

Luleå University of Technology

Department of Engineering Sciences and Mathematics
 High-temperature
Compression Strength of
High-alumina Refractory
Bricks Used in Rotary
Kilns of LKAB
Master’s Thesis in Materials Science and
Engineering
Dmitrij Ramanenka
September 2011

Supervised by:

Marta-Lena Antti, Division of Material Science

Luleå University of Technology

Department of Engineering Science and Mathematics

Division of Material Science

Abstract
Degradation of materials is one of the well-known challenges that are met in many industrial
situations. LKAB as one of the world’s leading producers of the upgraded iron-ore products
experiences varied degradation conditions during the processing of their products. One of their
central processes is the heat treatment of the products. Sintering, the final heat treatment step for
the products, is done in so called rotary-kilns. The interior of a rotary-kiln is isolated by a lining of
refractory bricks. The bricks inside the kiln are constantly affected by many different degrading
processes, where chemical attack together with thermal shock are considered to be the most
important reasons to severe degradation. This work is focused on the mechanical behavior of the
bricks and specifically their compression strength at room and service temperatures.

Three aluminosilicate-based bricks, belonging to the high-alumina category were tested. The bricks
are produced by Höganäs Bjuf AB and are marketed under the names: Victor HWM, Silox 60 and
Alex. Victor HWM contains approximately 79 wt% alumina (Al2O3), while Silox 60 and Alex contain
approximately 59 and 54 wt% alumina respectively. The primary mineral phase of Victor HWM is
corundum (α-Al2O3), Silox 60 is dominated by mullite (3Al2O3·2SiO2), while Alex is a blend of
corundum and mullite. The guideline value of room temperature compression strength is 80 MPa for
Victor HWM, 70 MPa for Silox 60 and 50 MPa for Alex.

Compression strength tests were performed at room temperature (RT) and in 700-1300 °C interval
with a variation of 100 °C between the tests. It was found that the strength increases with
temperature up to 900-1000 °C, where the peak of strength is noted for all the three materials. The
maximum strength increase, relative to the RT compression strength, of Victor HWM was found to be
approximately 65 %, for Silox 60 around 115 % and for Alex by more than 165 %. After passing 1000
°C, the strength decrease is relatively rapid for all the three materials, but is fastest in case of Alex
and is slowest in case of Victor HWM. At 1300 °C the compression strength decrease relative to RT
compression strength is approximately 63 % and 60 % for Victor HWM and Silox 60 respectively,
while the decrease is almost 86 % for Alex. Relatively to the bricks’ highest noted compression
strength, the strength decrease at 1300 °C is approximately 78 % and 81 % for Victor HWM and Silox
60 respectively, and around 95 % in case of Alex.

1
Acknowledgement
I would like to express my gratitude to my supervisor Marta-Lena Antti for the support during this
Master’s Thesis. Large appreciation to Stefan Adolfsson and LKAB for giving me the opportunity to
deal with this project. Furthermore, thanks to Erica Granberg and Höganäs Bjuf AB for technical
information and supply of the material. Thanks to Jesper Stjernberg at Material Science division at
LTU for sharing his experience with me. Moreover I am grateful to Johnny Grahn and Lars Frisk at
Material Science division at LTU for their help with the equipment. Thanks to Jan Granström at the
division of Mechanics of Solid Materials at LTU. At last but not least I am thankful to all the other
Master’s Theses students at the Material Science division for creating a pleasant working
atmosphere.

Dmitrij Ramanenka

2
Contents
Abstract ................................................................................................................................................... 1
Acknowledgement................................................................................................................................... 2
1 Introduction ..................................................................................................................................... 5
1.1 Background .............................................................................................................................. 5
1.1.1 Description of the problem ............................................................................................. 6
2 Objective.......................................................................................................................................... 6
3 Theoretical background................................................................................................................... 7
3.1 Ceramics .................................................................................................................................. 7
3.2 Refractory materials ................................................................................................................ 7
3.2.1 History ............................................................................................................................. 8
3.2.2 Manufacture .................................................................................................................... 9
3.2.3 Classification and description of refractory materials .................................................. 11
3.2.4 Mullite ........................................................................................................................... 14
3.3 Mechanical testing ................................................................................................................ 16
3.3.1 Cold crushing strength................................................................................................... 16
3.3.2 Compression strength at elevated temperatures ......................................................... 17
3.3.3 Brittle fracture in compression...................................................................................... 17
4 Scope of work ................................................................................................................................ 18
5 Materials........................................................................................................................................ 19
5.1 General .................................................................................................................................. 19
5.2 Notation and properties of the received materials............................................................... 21
6 Method and Experimental procedure ........................................................................................... 22
6.1 Method for determination of hot compression strength ..................................................... 22
6.1.1 Test specimens .............................................................................................................. 22
6.1.2 Procedure of the hot compression tests ....................................................................... 23
6.2 Preparation for the hot compression strength tests............................................................. 24
6.2.1 Test specimen preparation............................................................................................ 24
6.2.2 The equipment .............................................................................................................. 25
6.2.3 Manufacture of the ceramic support ............................................................................ 27
6.2.4 Planing of the ceramic rod ............................................................................................ 29
6.3 Analysis of contaminated Victor HWM ................................................................................. 29
7 Results and discussion ................................................................................................................... 31
7.1 Characterization of the materials .......................................................................................... 31

3
 7.1.1 X-ray diffractometry ...................................................................................................... 31
7.1.2 Microscopy .................................................................................................................... 32
7.2 High-temperature crushing tests .......................................................................................... 34
7.2.1 Compression strength of the bricks .............................................................................. 34
7.2.2 Fracture of bricks ........................................................................................................... 40
7.3 Evaluation of test method and test data............................................................................... 43
7.3.1 Effect of the powder...................................................................................................... 43
7.3.2 Double peaks ................................................................................................................. 44
7.4 Analysis of contaminated Victor HWM ................................................................................. 46
7.4.1 Analysis of raw materials............................................................................................... 46
7.4.2 Analysis of sintered materials ....................................................................................... 49
8 Summary and conclusions ............................................................................................................. 53
References ............................................................................................................................................. 54
Appendix A ............................................................................................................................................... i
Appendix B ...............................................................................................................................................ii
Appendix C...............................................................................................................................................iii
Appendix D .............................................................................................................................................. iv
Appendix E ................................................................................................................................................v
Appendix F ............................................................................................................................................... vi
Proportion calculation ......................................................................................................................... vi
Appendix G ............................................................................................................................................. vii

4
1 Introduction
Degradation of materials in industrial applications is a well-known phenomenon. Its influence can
appear in different ways. It can be a question about lowered production rates, quality of the product,
or safety, but it always diminishes the potential of an industrial process.

1.1 Background
LKAB, Luossavaara-Kiirunavaara AB, is one of the world’s leading producers of upgraded iron ore
products for the steel industry but also a supplier of mineral products to other industrial sectors.
Their main mineral resource is magnetite (Fe3O4) that is further processed into different suitable
products for sale on the market.

Iron ore pellets and sinterfines are the main products of LKAB. LKAB produces two types of pellets;
pellets for the blast-furnaces, which is also the most common reduction method of iron ore, and
pellets for the direct reduction. Pellets are circular, centimeter-large iron-reach solid pieces mainly
containing hematite (α-Fe2O3) that are produced after oxidation and sintering of magnetite together
with a binder and various additives (e.g. olivine, dolomite, limestone). The additives play an
important role in the behavior of pellets during reduction to smelted iron and that can for example
influence the stability of the reduction process and the quality of the end-product.
Sinterfines are iron-reach fines of either magnetite or hematite that are meant to be used in blast-
furnaces. There are also products which are mixtures of both pellets and sinterfines, all this
depending on the further iron ore reduction process used by the customer.

In order to oxidize and sinter their products, LKAB uses grate-kiln system at their production sites.
This process consists mainly of three parts; a grate band, a kiln and a cooling part. The kilns are rotary
kilns and the size varies between approximately 30-40 m in length and 5-7 m in diameter. These are
heated by a flame directed from one end of the kiln that is fueled by coal or oil. The band of grates is
a transport band for the raw material to the kiln. Furthermore, the system is made in such a way that
large part of the heat from the flame is not only warming the kiln, but is also used in order to oxidize
the raw material on the grates. This creates a temperature gradient along the whole system. The
end-part of the kiln, which is also the final step in the heat treatment, is the hottest spot in the chain,
while the beginning of the grate band is the coolest. The oxidation process starts already on the
grate-band, while sintering is finished in the rotary kiln. See Figure 1 for the illustration of the chain
process.

40 °C 400 °C 1000 °C 1120 °C 1250 °C

Oxidation and sintering

Grates Cooling
Rotary kiln

Figure 1. Illustration of the process in the grate-kiln system at LKAB.

5
1.1.1 Description of the problem
The interior of the rotary kiln is insulated by a lining of refractory bricks. Since the temperature varies
along the kiln different qualities of bricks are used in different temperature zones. In general, bricks
with higher alumina content are used in the warmest zones. The bricks inside the kiln are constantly
affected by different means, such as: high temperatures, harsh chemical environment, wear and
abrasion and very often all that simultaneously. However it is considered that chemical attack
together with thermal shock is the primary source to degradation of the bricks. All these conditions
make it unavoidable for the bricks to wear out. The best case scenario is when the bricks are wearing
out in a steady way. In average a certain lining of the bricks lasts approximately five years in the
rotary kilns of LKAB before it has to be replaced. Regular, planned service stops are made every 12-
15 month in order to repair the lining. However, unplanned emergency stops due to the fall-outs of
the refractory insulation are common and causes varied troubles to the company.

Degradation of the lining is influenced by a large number of factors: thermal, chemical, physical and
mechanical. Therefore there are many reasons to degradation, but also many ways to characterize a
refractory brick in order to classify the properties of the bricks. One of the most important and used
mechanical characterization methods is the CCS-test, cold crushing strength test, a uniaxial
compression strength test at room temperature (RT). This test is usually used as guidance to a brick’s
mechanical strength. Naturally, a brick with higher compression strength would mean a material that
is more resistant to mechanical stress. However, experience of LKAB has showed that cold crushing
tests do not always reflect the performance of the bricks. Nevertheless, the service temperature of
the bricks is far above room temperature. For example, recent emergency stops have suddenly
occurred at LKAB only few weeks after the installation of stronger bricks in a kiln. Because of the
short time the causes are mostly of mechanical character and not chemical. Experience from LKAB
has given rise to a couple of questions regarding the strength of the bricks, which will be also the aim
of this project.

2 Objective
The main aim of this study is to characterize compression strength at elevated temperatures of
refractories used at LKAB. An important aspect to investigate is how relevant it is to use cold crushing
strength measurements as guidance to the mechanical strength of a brick at its service temperatures.
Different questions will be considered. How does the strength vary over the temperature? Does
ranking of the strength at RT remain at elevated temperatures or do the differences disappear?
Simply meaning does the strongest brick at room temperature still remains the strongest brick at
elevated temperatures?

6
3 Theoretical background

3.1 Ceramics
Refractory bricks belong to the material category of ceramics. Ceramics are inorganic and non-
metallic materials. They are usually composed of metallic and non-metallic elements. The atomic
bonding within these materials ranges from purely ionic to totally covalent and can be a combination
of the two. [1]

Generally these compounds can be divided into three groups: traditional ceramics, refractories and
castables, and advanced ceramics. Stiffness and strengths of ceramics are in many cases analogous
with those of the metals. But in contrast to metals ceramics lack ductility, which makes them very
brittle and defect sensitive. [1]

3.2 Refractory materials

Refractories are materials which are usually used where the tolerance of high temperatures is
required. Most of the times they are used as a barrier between hot zones and relatively cold zones, in
order to protect the rest of a construction. Therefore, normally they function as thermal insulators,
but can also be manufactured and used for heat transport. The temperature tolerance is seldom the
sole requirement which has to be met in industry. Tolerance to thermal cycling and wear can be
some of the other requests. In other cases properties such as electrical resistance or electrical
conduction are needed. But probably the greatest challenge to refractories is the tolerance to
corrosive media. Especially considering that severity of corrosion attack is generally increasing with
elevated temperatures and in presence of abrasive/erosive conditions. [2] Refractories are typically
used as linings in furnaces for the refining of metals, but also in manufacture of glass and
metallurgical heat treatments.

Refractories are usually made of a complex combination of high melting crystalline oxides. Their
strength in compression is much higher than in tension and they have relatively high modulus of
elasticity, but as all ceramics they lack ductility. Refractories exhibit plastic deformation at high
temperatures, but mechanisms of it, as well as mechanisms of heat and electricity conduction are in
most cases not governed by the same principles as in metals. Due to the high variation of their
composition they can be designed in order to tolerate a wide variation of service conditions
(basic/acidic, salts/metals, liquids/gases). However, refractories are seldom totally immune to a
certain condition.

Refractories are rarely pore-free. Together with phase composition porosity is a very important
factor to consider during the manufacture of refractory products. Reduction in porosity increases
strength, load-bearing capacity and corrosion resistance. On the other hand typical characteristics of
refractories, such as thermal shock resistance and thermal insulation diminish with porosity
reduction. By balancing between porosity and phase compositions a good combination of properties
can be achieved for a specific service condition. [2]

7
3.2.1 History
The use of refractories can be traced to thousands of years ago, starting with the discovery of fire.
Unsurprisingly the first refractory materials were taken directly from nature. Easily accessible
materials such as stone and clay served as primary refractories for several thousands of years.
Accordingly, development of refractories over the years has been rather slow. On the other hand as
long as the primary source of the heat production was wood, the naturally found refractories
functioned relatively well since wood-fired ovens barely reached temperatures over 800 °C. About
4000 B.C. a new type of furnaces, convection-draft and forced-draft furnaces, made it possible to
reach temperatures around 1100 °C. This was also a starting point of the era of metals, but
nevertheless stone, clay and fired clay could still serve as adequate refractories. Not even when
porcelain was made in China by the year 600 A.D. and required temperatures up to 1350 °C, the need
of other refractories raised. Nor when the first types of blast furnaces in Europe by about 1400 A.D.
reached temperatures up to 1500 °C it was necessary- stone and fireclay were still acceptable, even
though the quality was relatively bad. One of the biggest reasons to this acceptance was that the
production rates were still relatively low. Probably the first trigger for the development of
refractories was the invention of Bessemer Converter in 1856 for the production of steel, a new type
of furnace. The furnace operated at temperatures over 1600 °C and under acidic corrosive
conditions. But the necessary refractories to withstand acidic attack were not yet developed. Just a
year after, another type of steelmaking furnace was invented, Siemens open hearth furnace. This
time reaching temperatures even higher than Bessemer Converter and operating with basic corrosive
slagging. Also the production capacity of open hearth furnace outstripped the capacity of Bessemer
Converter; therefore another requirement on the refractories became very important, namely their
durability in mass production industry. Compared to before there was now a need of a radical change
-consequently new types of refractories started to develop.

Silica (SiO2) brick was one of the new types of refractories that could withstand acidic slags, high
temperatures and wear better than traditional firebricks. In the beginning a lot of focus was
concentrated on the purity of one-element refractories. Silica brick was nevertheless not good
enough in all conditions; as one weakness it could not resist basic slag which could be found in the
open hearth furnaces. This opened door for the development of basic refractories. Magnesite
(MgCO3) was soon introduced into the linings of open hearth. Later during World War I dolomite
(MgO-CaO) bricks were developed and introduced, replacing magnesite brick. However as the
development progressed basic refractories such as magnesite and magnesite-chrome became the
refractory of choice in the open hearth furnaces. In the meantime, while radical changes happened in
the steelmaking industry, also other industries changed. Copper and aluminum smelting, glass and
many other industries were in need of new refractories, so the variations of magnesite brick were
developed. Later during World War II, super duty silica brick replaced the traditional silica brick, now
with higher refractoriness and longer life time. However after a while, basic and high-alumina
refractories became the preferred alternatives.

The development of the varied refractories during the first half of 20th century was significant, but
the quality was relatively low, partly due to the low demands during the two World Wars. After the
wartime a large demand raised on high-quality refractories which were specifically made for certain
service conditions. Especially when the three-phase arc furnace and the basic oxygen furnace (BOF)
in steelmaking were introduced, before 50s and in the end of 50s respectively, a new step in the
development of high-quality refractories was taken. [2], [3]

8
3.2.2 Manufacture

3.2.2.1 Raw materials

The choice of raw material for refractories is dependent on the application of the final product.
Basically two sources of raw materials can be distinguished: non-clay minerals and clay minerals.
Non-clays which are suitable for refractories include minerals such as: calcite (CaCO3) and aragonite
(CaCO3) which can be found in limestone; chromite (FeCr2O4), dolomite (CaMg(CO3)2), andalusite
(Al2SiO5), magnesite (MgCO3) and different feldspars (KAlSi3O8-NaAlSi3O8-CaAl2Si2O8). Also boehmite
(ϒ-AlO(OH)), diaspora (α- AlO(OH)) and gibbsite (Al(OH)3) which can be found in bauxite ore are some
other well-known non-clay minerals that are used in the production of refractories. Clays are
materials which have during the years been affected by different means, such as: water, temperature
cycles, abrasion etcetera. The grain size of these materials is generally smaller than that in the non-
clay group. Over the years and under various conditions the composition of these materials change;
for example by chemical decomposition, loss of water and/or acid-soluble compounds. Therefore
there is a wide variation of clay minerals, but all of them contain some chemically bonded water.
Kaolinite (Al2Si2O5(OH)4) is the most used clays in the production of refractories with ratio of 1:2 of
alumina to silica (Al2O3 : SiO2) content. [4]

Refractories are seldom one-component products, it is usually a mixture of compounds where one or
two components are dominating. Aluminosilicates are probably the most occurred compounds found
in refractory products, also small amounts of hematite (Fe2O3), rutile (TiO2), burnt lime (CaO) and
alkalis are some of the common constituents in refractories which naturally exist in the raw
materials. Additionally other additives can be used in the final material mixture in order to gain
specific properties.

3.2.2.2 Processing
After the raw materials have been extracted they need to be processed. This is usually firstly done by
crushing and grinding the material into smaller particles and after that followed by calcination.
Calcination is a heat treatment which is usually done well below the melting points of the
constituents in the materials. The purpose of this process is to bring thermal decomposition of the
material, sometimes also recrystallization and crystal growth. Some common examples of calcination
are calcination of hydroxides and carbonates, see (1) and (2) respectively for the reaction examples.

2Al(OH)3 → Al2O3 + 3H2O (g) (1)

CaCO3 → CaO + CO2 (g) (2)

Where (1) is dehydration of gibbsite and (2) is decarburization of calcite.

Other steps in the raw material processing can be size classification, analysis of the material
composition and drying.

Followed after the raw material processing is batching; this is made in order to correlate the defined
(by size and composition) raw materials to the desired final product. Simply meaning to find out the
needed mixture of materials for the final product. Mixing of the materials itself is an important step
for the quality of the final product; this can be done by dry or wet mixing. Processes following after
the mixing step are dependent on the type of products. Some of the products are unformed; then

9
they may be ready to be packaged directly after mixing, while others are further processed for
agglomeration or forming. However, most of the products are pre-shaped products, as in the case of
bricks. After mixing they are formed to a desired shape, this can be done in different ways such as:
uniaxial dry pressing, isostatic pressing, and extrusion or even by hand. The formed product, often
called green form, is then dried or cured. These products (green forms) can either be further heat
treated by firing them (burned products) or be left untreated (unburned products), followed by
eventual shaping and packaging. The firing is the process where the refractories obtain their
refractory properties and involves formation of ceramic bonds. However, at some stage even the
unburned refractories will create ceramic bond, usually after being installed. [2], [4] In Figure 2 can be
found a schematic of the manufacture process of refractories.

 Crushing/grinding
Raw material  Drying Unformed refractory-
 Calcining
processing end product.
 Size classification
 Composition analysis

 Shaping
Batching and mixing

Unburned formed
refractory-end
product.

Drying/curing  Forming
Green form
 Shaping

 Drying
Fused-cast refractory-  Arc melting
Burned formed
Firing end product.
refractory-end
Fused grains-further
product.
used.  Agglomerating
 Drying
Sintered grains-further  Firing
used.

Figure 2. Schematic over the manufacture process of refractory products.

10
3.2.3 Classification and description of refractory materials
The modern industry today involves a wide variation of service conditions and many of these
industries have processes that operate at elevated temperatures. As known, refractories are often
used in such conditions. Consequently there is a wide range of refractories available on the market.
The classification of the refractory materials is usually based on their raw material constituents, but
also on the methods of manufacture. Most of the refractories fall into the following main groups:
basic, high-alumina, silica, fireclay and insulating. There are also classes of “special refractories”;
silicon carbide, graphite, zircon, zirconia and others can be found here. As mentioned previously
most of the refractories are available in pre-formed shapes, such as bricks. But they can also be
supplied as clays, bonding mortars or ramming mixes, and others. [3]Due to the wide variation of
products, the content in this section will mostly refer to the refractories in form of bricks, since they
are the most common refractory products. In Table 1 an overview of a general classification of
refractory materials can be seen.

Ceramics

Traditional ceramics Refractories and castables Advanced ceramics

Basic High-alumina Silica Fireclay Insulating

% alumina:
Flux factor - Medium-duty Gr. No. _ Max bulk
Based on: - 50 - 85 - Superduty
- Type A density (g/cm3)
- deadburned MgCO3 or MgO - 60 - 90 - High-duty - Low-duty
- Type B From fireclay base:
- deadburned MgCO3 or MgO - 70 - 99 - Semi-silica
-16_ 0.54
+ chrome - 80 - 20 _ 0.64
- deadburned MgCO3 or MgO - 23_ 0.77
+ spinel - 26_0.86
- deadburned MgCO3 or MgO - Mullite brick
From high-alumina base:
+ carbon - Chemically-bonded brick
- Regular - 28_0.96
- Dolomite - Alumina-chrome brick
- Spall-resistant - 30 _ 1.09
... - Alumina-carbon brick
- Slag-resistant - 32 _ 1.52
...
- 33 _ 1.52

Table 1. General classification of ceramics and refractory materials.

3.2.3.1 Basic refractories

The name is derived from the fact that these refractories exhibit resistance to corrosion of basic
character at elevated temperatures. However, today this is not the whole truth, because some of the
basic refractories are capable of withstanding rather acidic conditions also. One such type is a direct-
bonded chrome-magnesite brick. Magnesia (MgO) is always present in this group of refractories,
therefore all the refractories rich in magnesia belongs to this group. But generally five categories can
be distinguished, products based on [3]:

 deadburned magnesite1 or magnesia

 deadburned magnesite or magnesia + chrome containing materials
 deadburned magnesite or magnesia + spinel
 deadburned magnesite or magnesia + carbon
 Dolomite (CaMg(CO3)2)

1
Deadburned magnesite (inert) is produced by sintering raw magnesite at a controlled temperature of 1750 °C.

11
Deadburned magnesite bricks have good resistance to basic slag and also to iron oxide and alkalis. By
combining deadburned magnesite with spinel a better resistance to spalling and lower coefficient of
thermal expansion can be achieved. In the early 60s it was succeeded to make direct bonded
magnesite-chrome brick, compared to earlier silicate-bonded brick, meaning that a thin film of
silicate minerals surrounded and bonded magnesite and chrome. With a direct-bond, a ceramic bond
between chrome and magnesite was achieved, resulting in products of high strength at elevated
temperatures and also exhibiting excellent slag and spalling resistance. Nevertheless chrome tends to
be replaced due to the environmental concerns. Another way of improvement is a magnesite-carbon
combination, resulting in increased corrosion resistance. This brick can sometimes replace the use of
magnesite-chrome brick. Another, widely used brick is a dolomitic brick and is a good low-cost
choice. [3]

3.2.3.2 Silica refractories

Silica refractories, also sometimes called acid refractories, are classified according to the amount of
impurities in the material by using so called “flux factor”, which per definition is equal to the
percentage of alumina plus twice the percentage of total alkalis. Additionally, some initial content
conditions have to be met by a product in order to be classified as silica brick. These are: Al 2O3
content less than 1.50 wt%, TiO2 content less than 0.20 wt%, FeO3 content less than 2.50 wt% and
CaO content less than 4.00 wt%. Also, average modulus of rupture should not be less than 3.45 MPa.
There are two types of silica brick [5]:

 Type A. Flux factor ≤ 0.50

 Type B. Flux Factor ≥ 0.50

Silica refractories have relatively high melting temperatures (1600-1725 °C) and they can withstand
relatively high pressures close to their melting points without failing. Accordingly their volume
constancy at elevated temperatures is also good. They have very good resistance to acid slags but
insufficient resistance to basic slags and iron oxides. Resistance to thermal spalling above 650 °C is
very good, while below 650 °C the thermal shock resistance is diminishing. Furthermore, silica
refractories are sensitive to impurities. [3]

3.2.3.3 Fireclay refractories

This type of refractories consists mainly of hydrated aluminosilicates, but small amounts of other
minerals are also present. One of the most common minerals used in the production of these
materials is kaolinite (2Al2O3·4SiO2·4H2O). [6]

Fireclay bricks are in contrast to the above mentioned groups of refractories not classified according
to their chemical content (except one class), but by their physical properties-maximum service
temperature and PCE2 (Pyrometric Cone Equivalence). The higher PCE the higher resistance of a
material to plastic deformation induced by temperature. They are divided into the following classes-
according to ASTM standard [7]:

 Low-duty fireclay (max. 870 °C; PCE 18-28)

 Medium duty fireclay (max. 1315 °C; PCE 29)

2
PCE-test is made by heating cone-shaped samples of the test material together with different standard cones
that are already defined by PCE-number. When the tip of the test cone reaches its base simultaneously as a
[3]
standard cone then the PCE-number is found.

12
  High-duty fireclay (max. 1480 °C; PCE 31½)
 Super-duty fireclay (max. 1480-1619 °C, PCE 33)
 Semi-silica fireclay (SiO2 content min. 72 %)

Among these classes high-duty and super-duty fireclays are subdivided into regular, spall-resistant
and slag-resistant fireclays.

In general, the melting temperature and therefore service temperature is increasing with higher
alumina content. Consequently the highest alumina content is in super-duty fireclays, with 40-44
wt% of alumina.

Low-duty firebrick is often used for backing up of other bricks of higher refractoriness while medium-
duty brick can often resist abrasion better than high-duty brick. [3] High-duty bricks have good
thermal-shock resistance and are widely used. Super-duty fireclay bricks have even better thermal
shock resistance and therefore also resistance to spalling.

Semi-silica bricks have some characteristics of silica refractories, such as excellent load-bearing
strength and volume stability at relatively high temperatures. [3]

3.2.3.4 Insulating brick

As the name reveals this products are used for insulation. Insulating brick can be made from different
types of oxides, and fireclay is one of the most widely used. The typical characteristics of these
products are light weight and low thermal conductivity thanks to the high porosity of these materials.
Therefore even classification is done according to their bulk densities, while the group numbers
multiplied by hundred reveals their nominal service temperature in degrees of Fahrenheit. According
to ASTM standard the groups are [8]:

 Based on fireclay, group No: 16 (845 °C); 20 (1065 °C); 23 (1230 °C); 26 (1400 °C)
 Based on high-alumina, group No: 28 (1510 °C); 30 (1620 °C); 32 (1730 °C); 33 (1790 °C)

Insulating bricks are mainly used as back-up materials since they do not tolerate abrasion and wear
very well. The advantages of these products can be faster heating of furnaces, lighter constructions
and lower heating costs.

3.2.3.5 High-alumina refractories

Refractories with alumina (Al2O3) content above 47.5 wt% are called high-alumina refractories; this is
according to ASTM standard and can therefore be somewhat different in other sources. The
classification within this group is primarily based on the alumina content in the material, resulting in
seven classes, as following [3], [7]:

 -50, -60, -70, -80 (± 2.5 % Al2O3)

 -85, -90 (± 2 % Al2O3)
 -99 (minimum 97 % of Al2O3)

Some other special classes of high-alumina refractories are mullite (e.g. 3Al2O3-2SiO2), chemically-
bonded, alumina-chrome and alumina-carbon products.

High-alumina bricks are some of the most used refractory products. Bauxite ore and andalusite are
the most common raw materials when producing high-alumina refractories. The dominating mineral

13
phases are usually either corundum or mullite. The refractoriness of these materials is in general a
function of alumina content, increasing with higher alumina content. Silica is the other main
component in this type of materials; therefore the slag resistance will be partly dependent on the
amount of it. High-alumina products which are high in silica will resist acid slags better than basic
slags. However, slag resistance is generally affected positively by increasing alumina content. Also
creep resistance is related to the alumina content. Since alumina has the highest melting point
among the main constituents creep resistance will be affected positively by increased alumina
content.

High-alumina products with alumina content between 70 wt% to 78 wt% where the dominating
mineral phase is mullite belong to the category of high-alumina mullite refractories. This is the
highest possible amount of alumina in “pure” mullite products. See Figure 3 for the binary phase
diagram of alumina-silica system. These products are often used where high purity and creep
resistance is required.

When high purity alumina and chromium oxide are fired together a solid solution bond develops in
the material. Alumina-chrome brick exhibit exceptional strength at a wide range of temperatures.
Additionally, the corrosion resistance is in many cases very good.

High-alumina chemically-bonded brick is usually made from bauxite in addition with soluble
phosphates, creating an aluminum-phosphate bond at 320-540 °C. But also creation of ceramic bond
is possible at higher temperatures. Phosphate bonded high-alumina bricks are often used in
aluminum industry due to their resistance to wetting and penetration by aluminum and its alloys.

Alumina-carbon bricks are bonded by carbonaceous bond created during heat treatment by special
thermosetting resins where graphite is usually the bonding agent. Naturally, these products are
often used in reducing environments. [3]

3.2.4 Mullite
Mullite is a solid solution phase of alumina and silica, regard Figure 3. It is a very important material
and is a major mineral phase in many ceramic products and is always present in clay products. The
general composition of mullite is Al4+2xSi2-2xO10-x where x denotes the number of missing oxygen
atoms per structural unit. Stable mullite exists in the range of 0.17≤ x ≤0.59 and up to a temperature
around 1850 °C. There are two mullite compositions which are known to be stable under normal
conditions. The first one is sintered-mullite (x= 0.25), 3Al2O3·2SiO2, also called 3/2-mullite, with
alumina content maximum 72 wt%. As the name denotes the production of mullite is done by
sintering-processes. The alumina content in this type of mullite will be governed by varied factors
such as sintering temperature, the duration of heat treatment, the initial composition of starting
material, particle size and others. The second type of stable mullite is fused-mullite (x=0.40) or 2/1-
mullite (2Al2O3·SiO2), alumina content maximum 78 wt%. This type of mullite is produced either by
melting (> 2000 °C) a raw material or by crystal growth techniques (Czochralski techniques). [10]

14
 [11]
Figure 3. Alumina-silica binary phase diagram.

Mullite exhibits high strength at high temperatures, low thermal expansion and therefore good
thermal shock resistance, and good chemical stability. See Table 2 where some of the properties of
mullite are compared to the properties of corundum.

Mullite (3Al2O3·2SiO2) Corundum (α-Al2O3)

Tm (°C) ≈1830 2044
ρ (g/cm3) ≈3.15 3.95-4.10
Linear thermal expansion (×10-6/°C)
20-1400 °C 3.5-5 5-10
Thermal conductivity (kcal/m/h/°C)
at RT 6 26
at 1400 °C 3 4
Compression strength (MPa)
at RT ≈1300 2600-3000
at 1400 °C [9] 800 250
Fracture toughness K1c, at RT
(MPa×m0.5) ≈2.5 ≈4.5

[12], [10]
Table 2. Properties of mullite and corundum.

Mullite is very rare in the earth crust; hence the formation of it in industry always involves processing
of other raw materials. Some of the most used starting materials for mullite formation in refractory
products are: bauxite ores together with silica; andalusite, kyanite and sillimanite; kaolinite and
pyrophilite. Kaolinite and pyrophilite are used in fireclay refractory products, among which kaolinite
is the most used. While the other above mentioned minerals are used in the production of high-
alumina refractories. [6],[10]

When kaolinite is fired it loses first its absorbed water at 100-200 °C and after that it dehydrates at
about 500-600 °C. Depending on the firing temperature different chemical changes happen. At 500-
600 °C, after the dehydration, metakaolin (2Al2O3·2SiO2) is formed. At 800 °C formation of one of the
following minerals takes place, all with the chemical formula Al2SiO5; andalusite, kyanite or
sillimanite. If firing is continued and reaches temperatures above 1000 °C, the formation of 3/2-

15
mullite occurs. The microstructure of fireclay products consists mainly of 3/2-mullite, cristobalite and
amorphous silica. [6]

Andalusite together with kyanite and sillimanite are the three polymorphs of Al2SiO5. The difference
lies in their crystal structures and that is dependent on temperature and pressure conditions. Among
these three forms andalusite is the less dense form, while kyanite is the densest. [13] The beginning of
transformation of kyanite, andalusite and sillimanite to 3/2-mullite occurs at 1150 °C, 1250 °C and
1300 °C respectively (this data is specifically for powders of grain size < 40 µm). [14] The
transformation takes place through the dissolution-precipitation process starting from the mineral
inclusions and the grain boundaries. Except mullite also cristobalite and amorphous silica are formed,
see (3). However, compared to materials of fired kaolinite these materials are richer in alumina and
therefore high-alumina refractories are the resulted products manufactured from them. [15]

3Al2SiO5 —> Al6Si2O13 + SiO2. (3)

Some of the characteristics of mullite that is formed from andalusite is its relatively low flux
(impurities) content and low porosity. Reason to low porosity is partly due to the fact that large part
of the amorphous silica that is formed during transformation is trapped in the capillaries of mullite
grains. Mullite formation from andalusite is stable and has low volume expansion, which is favorable
during manufacture of refractory products. [15]

When bauxite ore is used for formation of mullite it has first to be mixed with silica or silica rich
minerals. Then, when sintered, the formation of 3/2-mullite starts at about 1100 °C. This gives also
high-alumina refractories. The final microstructure of fired high-alumina bricks often consist of
mullite and/or α-Al2O3 (corundum) grains interlocked by the amorphous silica phase. [6]

3.3 Mechanical testing

3.3.1 Cold crushing strength
Cold crushing strength test or CCS-test is a uniaxial compression strength test (Figure 4) performed at
room temperature that is often used in order to characterize strength of refractory materials.

Sample

Figure 4. Illustration of a uniaxial compression strength test.

The test is performed until failure, meaning collapsing of the specimen or reduction of the specimen
height to 90 % of its original value. Highest applied load is noted and the compression strength is
calculated by:

16
 (4)

Where σc (MPa) is compression strength of the material, F (N) is the maximum noted applied load
during the test; A (mm2) is average of the areas of the top and bottom of the specimen perpendicular
to the applied load direction. [16]

3.3.2 Compression strength at elevated temperatures

Compression strength tests at high-temperatures are rare; no actual standards exist for this kind of
test. It requires non-ordinary equipment and are very time consuming. Normally if high-temperature
strength is of interest then the closest alternative to compression tests are so called hot MOR
(modulus of rupture) tests (bending test). This requires also special equipment and is time
consuming, but the maximum applied loads are much lower compared to compression tests.

The general behavior of the materials at elevated temperatures, as known, is that the strength
diminishes. However, there are cases when this is not true; the strength can not only remain but also
augment with temperature. See Figure 5 where hot MOR of two refractory materials is presented.

[17]
Figure 5. Dependence of bending strength on temperature for two refractory materials.

3.3.3 Brittle fracture in compression

It has been shown that materials such as bricks, concrete, rocks and even glass behave very similar
when regarding their fracture modes. There are many evidence confirming that the dominating crack
grows in these materials is of mode I type (cleavage mode). This is the same mode that occurs in pure
tension. Therefore, propagation of cleavage cracks is the most noticed event during a compression
test. However, many authors refer also to an overall mode occurring in compression, so called
splitting-mode that cannot be clearly referred to the classical mode I, II or III. [18] In any case,
separation of the material and formation of opened visible cracks is a fact. Not surprisingly different
kind of flows (cracks, pores, voids…) will function as stress raisers in the materials and the crack
growth will originate from them. [19] It has been found that in a material such as granite most of the
cracks are initiating at the grain boundaries. More specifically those grain boundaries that separate
quartz grains are extra prone to crack initiation. Also, in general, the grain boundaries that run 20-40°
from the applied load direction were found to be most prompt to start new cracks. [20]

17
4 Scope of work
In this work room temperature and high temperature compression (uniaxial) strength tests of three
different materials were performed. Scanning electron microscope (SEM), electron dispersive
spectroscopy (EDS) and x-ray diffractometry (XRD) were used as characterization methods. Mineral
phases of the as-received materials were analyzed. Morphology of the bricks was studied. Phase
transformation and strength variation after a long soaking at elevated temperature was detected.
Deformation modes at different temperatures were investigated. Two raw materials for one of the
materials were analyzed and a contaminated brick-sample made from one of the raw-materials was
studied.

18
5 Materials

5.1 General
Höganäs Bjuf AB is the supplier of the refractory bricks to LKAB that are used in their rotary kilns.
Three qualities (products) were considered in this study. Their trade names are:

 Victor HWM
 Silox 60
 Alex

Most of the focus was aimed to Victor HWM and Silox 60, while Alex was tested for the sake of
comparison. Below can be found light microscope images of the three brick types.

a) 1mm b) 1mm

c) 1mm d) 1mm

Figure 6. Light microscope images (x20) of Victor HWM (a), Silox 60 (b) and Alex (c). d) Represents Viktor HWM (x15)
from a different batch compared to a).

These are all aluminosilicate-based bricks, belonging to the category of high-alumina bricks, but that
have been manufactured from different raw materials. Victor HWM is based on bauxite, Silox 60 is
based on andalusite and Alex is based on chamotte reinforced by bauxite. They all have the same
main chemical constituents (preferred metal oxides) but in different proportions (note these are not
their crystalline phase composition). See Chart 1 and Table 3 for the chemical analysis and guideline
values of the properties for these three bricks.

19
 1% 0.3%
Victor HWM 0.9% Silox 60 Alex
2.2% 1.4%
0.2% 0.1% 1.3%
0.5% 1.5% 0.5% 2.1%

17%
37%
40%
59% 54%
79%

Al2O3 SiO2 TiO2 Fe2O3 CaO Alkalies

Chart 1. Chemical analysis (wt %) of Victor HWM, Silox 60 and Alex representing the abundance of the most preferred
oxides.

Victor HWM Silox 60 Alex

3
Bulk density (BD) (g/cm ) 2.70 (2.65-2.75) 2.45 (2.40-2.50) 2.32 (2.25-2.40)
Apparent porosity (AP) (%) 18 (17-21) 17 (15-19) 19 (17-21)
Cold crushing strength (CCS)(MPa) 80 (60-100) 70 (50-90) 50 (30-70)
Refractoriness under load
-T0.5 1400 1500 1380
-T5 1590 1650 1470
Thermal shock resistance >30 cycles >30 cycles >30 cycles
Linear thermal expansion at 1000
°C (%) 0.6 0.6 0.6
Thermal conductivity (W/(m·K))
Measured at:
-500 °C 1.8 1.2 1.8
-750 °C 1.9 1.4 1.6
-1000 °C 2.1 1.9 1.5
-1250 °C 2.0 2.0 1.6

Table 3. Guideline values of the properties of Victor HWM, Silox 60 and Alex. Tests according to DIN-standards.

In Figure 7 can be found QemScan images of the polished surfaces of the three as fabricated
materials representing distribution of the mineral phases.

a) 2 mm b) 2 mm c) 2 mm

Corundum Mullite Mullite (low Al) Andalusite

[21]
Figure 7a-b. QemScan images of Victor HWM (a), Silox 60 (b) and Alex (c).

20
All the three types of bricks are dry-pressed at approximately 60-100 MPa while the water content is
held at 2-3 wt % during the compression state. The burning of the bricks takes place in a tunnel kiln
at 1350-1360 °C.

These materials are highly heterogeneous and as can be seen in the table above a property such as
cold crushing strength varies significantly within the same product type. One of the main reasons to
the variation in properties lies in the manufacture of the bricks and more precisely during their
sintering process. When the green bodies are fired in the tunnel kiln they are stacked one on
another, as can be seen in Figure 7. The geometry of the kiln and the location of the burners in the
kiln create an uneven heating of the heat chamber. Due to this there is a temperature difference of
approximately 100 °C, at the most, between the bottom and the upper part of the kiln - hence a
variation in compression strength appears due to these dissimilar burning conditions.

Figure 8. Picture of stacked refractory bricks when sintered in a tunnel kiln.

5.2 Notation and properties of the received materials

In Table 4 indication of the properties on the received samples from Höganäs Bjuf AB can be found.
Every sample notation represents samples from the same batch.

Quality CCS (Mpa) BD (g/cm3) AP (%) Sample notation

Viktor HWM 73,3 2,71 19,4 1-5v

Victor HWM 66,3 2,73 20,4 9-16v
Victor HWM 97,9 2,86 17,2 17-24v
Victor HWM 61,6 2,72 19,6 35-40v
Silox 60 66,9 2,46 17,9 1-5s'
Silox 60 56,6 2,41 19,1 1-8s''
Silox 60 50,7 2,44 19,2 25-29s
Silox 60 66,2 2,46 18,5 30-34s
Silox 60 52,9 2,5 18,8 47-54s
Alex 45,1 2,35 17,1 41-46a

Table 4. Specifications and notation of the received samples. (Measured by Höganäs Bjuf AB)

21
6 Method and Experimental procedure

6.1 Method for determination of hot compression strength

No standards for hot crushing tests of refractory products were found to exist. In order to find a
proper methodology - standards for cold crushing tests, hot MOR tests and recommendations from
other works were considered.

6.1.1 Test specimens

According to ASTM C133, standard test method for cold crushing strength of refractories, the
specimens from bricks of density above 1.60 g/cm3 are recommended to be in shape of 51 mm-
cubes or cylinders of 51×51 mm dimension. The height should be parallel to the original direction of
pressing of the brick.

European standard EN 993-5, method for the determination of cold crushing strength of dense
shaped refractories, recommends test pieces of either cylinders (50±0.5 × 50±0.5 mm) or cubes
(50±0.5 mm). It is also said that if it is not possible to obtain this size from the test item, then
cylinders 36±0.3 mm in diameter and 36±0.3 mm in height can be used. Furthermore, the top and
the bottom ends of the test pieces should be flat and also parallel, with height difference not more
than 0.2 mm between the load-bearing faces.

In this study, the tests were limited by the existing equipment; the available frame was limited to 100
kN loads. Therefore the recommended large specimen sizes described above could not be used,
which would have required crushing loads by estimate up to 200 kN. Even though it was suspected
that the compression strength of the tested refractories would considerably decrease at elevated
temperatures- it was considered necessary to diminish the recommended specimen size. It was
decided to take the maximum room temperature compression strength of Victor HWM (100 MPa) as
a required upper limit. By calculating (Areamax=Forcemax/Strengthmax) it was found that the area of the
samples could be maximum 1000 mm2. That is a cylindrical sample of approximately 35.7 mm in
diameter or a cube with a side length of approximately 31.6 mm. The size of the cylinder would be in
the vicinities of the acceptable size by EN 993-5 standard, described above. Fortunately, it was also
found out that Höganäs Bjuf AB had possibilities of drilling cylindrical samples of 35 mm in diameter.
Therefore, for the sake of simplicity it was chosen to prepare samples of cylindrical shape with
diameter of 35 mm and height of 35 mm, as a slight modification to the European standard EN 993-5,
see Figure 9.

35 mm

// 0.2 mm 35 mm

Figure 9. Illustration of specimen’s dimensions.

22
6.1.2 Procedure of the hot compression tests
ASTM C 583 standard, test method for modulus of rupture of refractory materials at elevated
temperatures, was considered in order to find appropriate heating procedure for the test specimens.
The standard quotes that the rate of heating of burned refractory products shall not exceed 330 °C/h
from room temperature up to 980 °C, and shall not exceed 110 °C/h from 980 °C up to the test
temperature. The test temperature should be maintained for a minimum of 3 h.

In case of this study the heating rate was set to 300 °C/h up to 1000 °C and from 1000 °C to the test
temperature the heating rate was set to 150 °C/h. This is slightly above the recommended heat rate,
but it was believed that this would not have any significant influence in this case since the materials
and equipment can tolerate that. The test temperature was maintained for 2 h, with respect to a
three times smaller volume than for the specimens in the standard described above. The cooling rate
was set to 150 °C/h.

Rotary kilns of LKAB can experience local temperatures up to 1350 °C, but vary along the kiln. Also,
naturally the temperature of the bricks varies along the thickness too. In order to find relevant test-
temperatures tests were firstly performed at 1200 °C, followed by increasing or decreasing the next
test-temperature by 100 °C. Step by step a following test-temperature range and sequence was
found to be reasonable and of interest: 700-900-1000-1100-1200-1300 °C.

ASTM C133 standard, standard test method for cold crushing strength of refractories, recommends
loading rates of either 12 MPa/min, 31.2 kN/min or 1.3 mm/min for bricks with density over 1.60
g/cm3. Loading at a constant stress rate is however preferable if possible.

The compression rate for the tests was chosen to be 1 mm/min for the whole range of test
temperatures. There is a certain possibility that the materials exhibit some visco-plastic behavior at
temperatures over 1000 °C, meaning that the strength determination is also load-rate affected,
however the chosen compression rate is probably relevant even in that case.

ASTM C133 standard also highly recommends usage of a spherical load bearing that can distribute
load evenly if the samples are not perfectly parallel. In practice such bearing at planned test
temperatures was found to be difficult and expensive to implement. It has to be of a heat tolerant
material and has to be specially manufactured, therefore expensive. It has to be relatively large (large
curvature radius) in order to function well and it has to be quite massive, since the loads on it would
be partly of the bending character. Attempts to manufacture such bearing in the facilities of the
university were done but all unsuccessfully. However, it was important to find a replacement for the
bearing in order to avoid possible bending stresses on the sensitive equipment (pushing bars) during
tests. Aluminum oxide powder was found to be a relevant alternative that could be used under the
samples in order to adjust the differences in parallelism. The powder was evenly placed under the
samples with initial thickness of approximately 5 mm, after the applied compression the thickness
was not more than 1 mm.

A pre-load during the heating up of the samples was necessary. This is required in order to compress
the alumina powder below the sample but also simply in order to keep the contact with the sample
and keep the sample steady during the heating. There exist no recommendations about the applied
pre-load. ASTM C133 states that for high density refractory samples it is possible to increase load as
fast as it is wished up to 50 % of the expected maximum strength and then use the recommended

23
compression rate. This implies that high-density refractories are relatively insensitive to loads. It was
chosen to set the pre-load of the tests to 1 kN during the heating, which is 1-2 % of the maximum
room temperature strength of the test-bricks. The same pre-load was applied in the whole
temperature interval. In reality the pre-load was somewhat lower during the heating, since the load
cell was affected by the heat and measured approximately 300 N above the real values.

6.1.2.1 Summary-test procedure

 Test temperatures: RT, 700-900-1000-1100-1200-1300 °C.
 Pre-load: 1 kN.
 Heating rate: 300 h/°C up to 1000 °C. 150 h/°C after 1000 °C.
 Compression rate: 1mm/min.
 Cooling rate: 150 h/°C.

6.2 Preparation for the hot compression strength tests

6.2.1 Test specimen preparation
Samples were received from Höganäs Bjuf AB in form of cylindrical rods, with diameter of 35 mm and
length of approximately 60 mm. The samples were drilled out from the refractory bricks that already
passed the sintering process. In the best cases maximum eight samples could be drilled out from one
brick. Only the samples from the same brick were considered as the samples from the same batch.
The drilled, long side of the cylinders was smooth and no preparation of that was considered to be
necessary. A couple of methods were tried out for the height, planing and parallelism preparing. It
was found that simple cutting by the facilities of university was not reliable as a method, due to bad
parallelism. Therefore, also grinding of the samples was made. The cutting of the samples was
performed in Discotom 2 (Struers) cutting machine using a diamond cutting wheel and water cooling.
The force during the cutting was applied manually and one cut engaged less than a minute. Polishing
was done in Phoenix 4000 (Buehler) semi-automatic polishing machine. The samples were fixed in a
holder at the long sides, while the sides to be polished were on the same level relative to each other.
Three samples, 120 ° between each other, were grinded at a time. A coarse diamond grinding disc
was used together with water cooling, grinding direction was clockwise (150 rpm/min) and only very
little pressure was applied. As a matter of fact, the pressure was so low that the machine registered
zero applied pressure; this is of course not true in the reality. Due to the samples’ nature, the grains
in the material can easily detach if too much pressure during coarse grinding is applied. The grinding
rate was approximately 1 mm in 20 min by the applied pressure.

Cutting was firstly done on one side and then polished until the perpendicularity was satisfied. That
was checked by a bevel protractor, this can be done without taking samples out from the holder. If
necessary the samples have to be re-fixed in the holder. It required some degree of practice before
getting the drilled sides of the samples parallel to the walls of the holder - which is of coarse
necessary in order to get the grinded sides perpendicular to the drilled sides. When the
perpendicularity was satisfied the second cut was made approximately 36-37 mm from the other
end. After approximately 10 minutes of grinding the parallelism of the samples was checked by
caliper. If satisfying then the samples were grinded until approximately 35 mm. If parallelism is not
satisfying the samples have to be re-fixed in the holder. “Dummies” were used in order to replace
samples that were finished before others.

24
All the samples prepared in this study had load-bearing sides parallel to each other with height
difference of not more than 0.2 mm. Some of the samples were not successfully prepared and were
therefore not accepted. Most of the time this was due to heavy detaching of the material during
polishing. Victor HWM (9-16v) in the middle-strength category was specifically prone to that; half of
the samples in this batch were excluded. Figure 9 shows a final prepared sample.

Figure 9. Picture of a prepared sample.

6.2.2 The equipment

For the performance of the warm compression tests a furnace (SSL) and a compression testing
machine were available. The furnace was equipped with water cooling system and can according to
the manufacturer be used up to 1600 °C. The compression testing machine was fitted with 100 kN-
load cell and the frame was limited to loads of maximum 100 kN. The heat-tolerant compression
equipment was made of aluminum oxide that was fixed into water cooled steel supports. Detailed
description of the equipment and the preparation of it can be found further down.

6.2.2.1 The furnace

The furnace is made of two semi-circular parts that are fixed on the frame of the compression
machine. When the furnace is in use the two parts are closed together creating a cylinder-like
heating chamber (useable space: D=120, h=260) that is warmed up by six MoSi2 heating elements.
Two of the heating elements had to be replaced prior the tests. Two entrances opposite to each
other are running parallel to the furnace’s long side and are centered over the heating chamber.
These entrances are used for the fitting of the desired equipment. They were broadened to 48 mm in
diameter in order to fit the used equipment, which is also the maximum possible diameter on the
upper part of the furnace. Parts of the insulation in the furnace had to be repaired; this was done by
a high-temperature spackling paste “Kleber” and a hardener.

The two parts of the furnace have to be well centered to the centrum of the machine and also fit well
together. High temperature cotton was used to seal the existing openings. If the two halves of the
furnace are not fitted well together the heat flow from the furnace can be significant, especially
above the furnace where the load cell is situated. This creates a risk of damaging the load cell but
also the cell starts to measure incorrectly due to its own abnormal thermal expansion. A steel sheet
with an on-glued insulating layer (see Appendix A) was placed on the top of the furnace to block the
heat flow additionally. Furthermore, a ventilation output was placed close to the top of the furnace.
Power supply cabling for the heating elements is situated on top of the furnace, when the steel sheet
is placed on top of the furnace the cabling is subjected to higher temperatures than usually.
However, it showed that the blockage sheet could insulate from the heat flow without causing any
troubles with the cablings below it.

25
The power control box of the furnace was not connected to the furnace and that had to be installed.
The maximum output power supply can be regulated by a switch inside the power control box. This
should be checked out in order to be sure that the output power is not limiting the possibility of
reaching the desired temperature. The output power switch can be used when no automatic
temperature control system exists, which was not in this this case.

The temperature control system (Eurotherm) of the furnace was equipped with a limited number of
control functions. The ramp-function was available, but no dwell or step-function software was
installed. Temperature measurement was done by two thermocouples placed close to the sample.

Prior to the warm tests the water cooling was tested during a long run (>12h) of the furnace at 1200
°C and it was confirmed to function well by regularly checking the output water temperature. It is
recommended to do regular controls of output water temperature to be sure that the flow is
sufficient and free.

6.2.2.2 The compression equipment

The design of the heat-tolerant equipment was planned to be in form of pushing bars that could be
fitted to the furnace.

Due to the high test temperatures a ceramic material was chosen as heat-tolerant equipment.
Alumina as one of the cheapest and well-performing options was the candidate in this case. At room
temperature high purity alumina (> 96 %) exhibits compression strength of around 2000-2700 MPa.
[12]
This is significantly above the required around 100 MPa at the room temperature tests of the
refractory samples. Maximum service temperature of high-purity alumina (>96%) lies generally
around 1550-1700 °C without applied load. [12] According to [9] the compression strength of 99.8 %
alumina at 1400 °C is approximately 250 MPa. At the planned test conditions (max 1300 °C) the
compression strength of alumina should of course be higher, but it is difficult to say to what degree.
In any case, since the frame is limited to 100 kN loads, the maximum possible pressure on the
equipment is around 100 MPa independent on the test-temperature.

The bottom part of the high-temperature equipment (alumina tube + steel support) was available.
The dimensions of it were acceptable, the equipment was firstly found to be sufficiently strong for
the planned tests. However, after approximately ten tests the steel support for the bottom alumina
tube started to buckle. This was solved by installing an extra support under the alumina tube that
could unburden the primary steel support. Also the whole bottom part had to be re-aligned, in this
case simply by placing steel sheets between the steel support and the rig.

For the upper part of compression equipment solid alumina bar (Frialit F99.7) fixed into a water
cooled steel support was designed and used.

With total length of approximately 70 cm the pushing equipment could be easily affected by bending
stresses if not properly aligned, which was also one of the challenges.

The supply of water to the furnace and compression equipment was not preinstalled; therefore a
new main water connection had to be set up. This was done by splitting an already existing water
junction. Also some of the old existing hoses and hose links had to be repaired due to leakage. For
the illustration of the equipment set-up see Figure 10. Additional pictures of the equipment can be
found in Appendix A.

26
 4

3
H2O

6
1
H2O
2
≈70 cm

1
3 H2O

Figure 10. Illustration of the equipment set-up. Notations: 1: Al2O3-pushing bars; 2: sample; 3: water-cooled steel
supports; 4: load cell; 5: thermocouples; 6: heating elements.

6.2.3 Manufacture of the ceramic support

Ceramic support is a part made of alumina that was placed between the sample and the lower
pushing tube, see Appendix A, Figure 1. Due to an accident the existed ceramic support had to be
replaced. In order to save time it was chosen to manufacture such ceramic support in the facilities of
the university. The material of choice was alumina (Al2O3); the specifications of the powder material
can be seen in Table 5.

Material Producer Product name Particle size (µm)

Al2O3 Baikowski (France) CR10 <0.45

Table 5. Specifications of the material used for the manufacture of the ceramic support.

Firstly the powder of alumina had to be dried; this was done in a drying furnace at 110 °C over the
night. The next step is packaging of the powder into a desired form. At this step it is important to
take into the account the shrinkage of the powder during the after-following steps. The alumina
powder was packaged into a stiff rubber tube (Dinner=73 mm) which was sealed in one of the ends.
The powder was packed by hand as much as possible reaching a total height of approximately 75
mm. Around the rubber tube a tight metal tube was fitted in order to support the inner rubber tube
in a following pre-compression of the powder, which would otherwise expand radially. The
compression of the powder was firstly made in a uni-axial press using a maximum pressure of
approximately 60 MPa during several minutes. After this forming step the sample is roughly half of
the starting height and the diameter is slightly larger. Afterwards the rubber tube was pressed out
from the metal tube without taking out the sample and the remained free end was sealed. The

27
sealing was done by using steel plugs together with vacuum grease fixed by isolation tape. The final
compression was done in a Cold Isostatic Pressure (CIP) machine, using pressure of approximately
350 MPa and holding it for a couple of minutes. Afterwards the sealings of the rubber tube were
removed and the sample was carefully collected out from the rubber tube.

The manufacture of the ceramic support had some challenges to overcome. It was found to be
important to pre-compress the samples prior the CIPing. If the powder is pre-compressed then most
of the air in the powder is gone and the plastic tube has to deform in a much lesser extent when
CIPed, avoiding that the tube penetrates the powder or the sealing starts to leak during CIPing.
However, the pre-compression was not always succeeded since the samples had a tendency to split
in slices perpendicular to the applied force. It was found to work out best when the pressure was
held for several minutes and not exceeding 60 MPa.

After the compression of the sample the next steps in the procedure are the heat treatments. This
implied firstly drying of the sample at 600 °C during 5 hours (heating rate 1 °C/min, cooling rate 2
°C/min) in order to vaporize bonded water and burn out potential hydrocarbons. The obtained green
body is sensitive to all kinds of stresses and should be handled very carefully. In order to make the
final grinding and polishing easier it is advantageous to make most of the planing and grinding of the
sample already at this step, before the sintering. The disadvantage is the risk of the introduction of
flaws into the material. However, in case of the grinding of the green body this was done on a P60
grit SiC-paper. To make sure the plane sides are approximately parallel a “watch” was used. Firstly
one side is made flat and a parallel line to that side is drawn closely to the opposite side, therefore
the line functions as a reference line during planing, see Figure 11. After planing and grinding the
green body was later on sintered at 1575 °C during 4 hours (heating rate 3 °C/min, cooling rate 2
°C/min). After sintering the alumina sample has a hardness of approximately 2000 HV. Fortunately if
most of the grinding and planning is already done at the green body step then only the final planing
and polishing is left. A “watch“ can again be used for drawing a reference line to facilitate grinding.
Diamond grinding disc with 55 µm grits was used for planing. If larger volumes of the material had to
be removed then a coarser diamond grinding disc was used firstly. It is important to be careful when
using coarse diamond grinding since it can easily introduce damage to the surface if too much
pressure is applied. However, by this method and by frequent controls of parallelism it is possible to
achieve good tolerances. The final polishing was done by the following sequence: 9 µm-3 µm-1 µm
diamond grits, creating a mirror-like surface. The used alumina support in this case had a thickness of
≈23 mm and parallelism variation of ± 0.1 mm of the flat sides (load-sides). The accuracy of the sides
around is not of importance. See Appendix B where calculations are made for the approximate
minimum required thickness of the ceramic support, based on the bending strength of alumina.

//

Figure 11. Illustration of the ”watch”.

28
6.2.4 Planing of the ceramic rod
The delivered alumina bar for the upper compression equipment was found to be unacceptably
oblique and beyond the tolerances of the manufacturer itself. The diameter was found to be almost
3 mm larger than expected. Because of this the steel holder for the ceramic bar had to be turned
again in order to fit the bar. The flat sides of the bar were not perpendicular to the long side and
therefore the long side was out of the perpendicularity line at the most as much as 6 mm, see
Appendix A.

It was decided to regrind the flat sides of the bar and make them parallel rather than waiting for the
correct delivery.

In order to regrind the alumina bar it was fixed by two clamps for the sake of stability while the long
side of the bar was oriented perpendicular to a diamond grinding disk by a bevel protractor, see
Figure 12. The first grinding step involved a coarse grinding followed by a 55 µm diamond grit
grinding. The force was applied by hand; due to the length of the bar it is important to be steady,
otherwise even small deviations will give bad result. By applying very little force and doing frequent
controls it is possible to obtain acceptable perpendicularity. After planing the bar was deviating from
the perpendicularity line at the most approximately 1 mm. The final polishing was done on master
tech polishing cloth with a sequence of 9, 3 and 1-µm diamond grids.

Figure 12. Illustration of the re-grinding of the alumina bar.

6.3 Analysis of contaminated Victor HWM

Höganäs Bjuf AB experienced problems during the production of Victor HWM. It seemed to be
contaminated; an assay to trace this contamination was performed in this work. (Note: Victor HWM
samples tested in compression strength tests were not contaminated).

Two base-raw materials called New RASC and Old RASC were received from Höganäs Bjuf AB (but are
not manufactured by them). New RASC is the one that was suspected to be contaminated, while Old
RASC was the “normal” base-material. The two raw material types were in form of fragments of
varied sizes, the largest particles up to one to two centimeters large. These were crushed into fine
powders in a mortar. EDS- and XRD- analysis were performed on the powders.

29
Also a sample of contaminated Victor HWM which was sintered from New RASC base was received.
EDS-analyses on the surface and suspected contamination particles were performed.

30
7 Results and discussion

7.1 Characterization of the materials

7.1.1 X-ray diffractometry
Four main mineral phases are present in the bricks, these are: Corundum (Al2O3), Mullite
(3Al2O3:2SiO2), Andalusite (Al2O3:SiO2) and β-cristoballite (SiO2). Also traces of β-quartz (SiO2) exist in
Alex. As can be read in Graph 1 Victor HWM is dominated by corundum, Silox 60 consists of mullite,
but has also significant amount of andalusite; and Alex is a mixture of different phases. However, due
to the heterogeneity of the materials, the diffractograms should be considered only as an
approximate guidance to the content.

3
Graph 1. XRD-analysis of Victor HWM, Silox 60 and Alex . (Range: 10-90 2θ degrees. Step width: 0.05 degrees.
Time/step: 5 sec.)

The diffractograms reveal also a clear detection of amorphous peak in case of Silox 60 and Alex.
According to the information from Höganäs Bjuf AB the glass content in Victor HWM is 8-10 wt%, in

3
Values in the diffractogram of Alex were collected from a personal contact (Jesper Stjernberg, LTU).

31
Silox 60 14-16 wt% and 19-20 wt% in Alex. No clear amorphous peak is visible in the presented
diffractogram of Victor HWM; however, this is probably specifically for the tested brick that was from
a high-strength category.

7.1.2 Microscopy
Morphology of the bricks is rather similar. Different minerals of different grain sizes are blended and
distributed randomly in the materials. Matrix of smaller grains is surrounding the larger grains.
Unburned andalusite is of relatively small grain size and is distributed in the matrix. See also Figure
13 for a representative image of the morphology.

60
55
50
45 O
3 2
40
Al

Atomic %
1 35
30 Si
25
20 Ti
15
10
Fe
5 0,22 2,02 K
0,47 0,11
4 0
1 2 3 4
Spot number

Figure 13 and Chart 2. (left) SEM image (x140) of Silox 60 representing typical morphology of the materials. (right) EDS-
analysis of the specified spots.

Consider EDS-analysis in Chart 2. The marked large grain in the centrum of Figure 13 is a corundum
grain with high amount of titanium oxide and iron oxide. Large amount of the materials’ rutile and
hematite is solved in corundum. In the presented case the matrix is mostly consisting of fine mullite
grains. Potassium oxide tends to mostly exist in the matrix, which is probably solved in the glassy
phase that is present in the matrix. The grain size in the materials can be up to 5 mm. Up to 20-25 %
of the grains are in the range of 1-5 mm large, while the rest is below 1 mm. However, this is of
course at the macroscopic level, in reality the grains are lumps of smaller grains. A clear example can
be seen in Figure 14a-b, 14a is a magnification of the black dashed area in Figure 13. Also, rutile-rich
matrix can be clearly seen between the grains, see Figure 14b.

Al2O3-grains

TiO2-rich matrix

a) b)

Figure 14a-b. SEM images of corundum. a) Silox 60 (x1200, dashed area in Figure 13). b) Victor HWM (x1800).

32
When andalusite is transformed into mullite excess silica is released as amorphous silica.
Approximately 20% amorphous silica is formed from an andalusite monocrystal. Some of it will be
expelled, but most of it is trapped in the mullite grains as “islands” and in capillaries, see Figure 15
and Figure 16. The amorphous silica can further react with excess fine corundum, if such exists, and
form secondary mullite that can additionally enhance the strength and corrosion resistance of the
material. [15]

60
55
50
45
O
40 Al
3

Atomic %
35
30 Si
25
2 Ti
20
15
1 Fe
10 0,09 0,06
0,59
5 0,78 1,31 1,54 K
0,63 1,27 0,3
0
1 2 3
Spot number

Figure 15 and Chart 3. (left) SEM-image (x550) of Si-rich areas in the mullite grains (Silox 60). (right) EDS-analysis of the
specified spots.

SiO2-rich capillaries
Mullite

a) b)

Figure 16a-b. SEM images of Si-rich regions in the mullitized andalusite grains (Silox 60). a) x2000. b) x2500.

33
7.2 High-temperature crushing tests
7.2.1 Compression strength of the bricks

7.2.1.1 Normalized compression strength- general trends

General trend of hot compression strength (HCS) over the temperature interval 700-1300 °C for
Victor HWM, Silox 60 and Alex can be seen in Chart 4a-d. The graphs represent percental change of
compression strength obtained at elevated temperatures relative to the bricks’ room temperature
compression strength. Colored signs denote tests made on samples from the same batch, values
between parentheses are the brick’s room temperature compression strength, while the lines
indicate the general trend.

a. b.

c. d.

Chart 4a-d. Normalized values of hot crushing tests of Victor HWM (a), Silox 60 (b) and Alex (c) denoting the general
trend of compression strength over the 700-1300°C interval.

The compression strength is significantly influenced by the temperature. The general trend of the
compression strength over the temperature interval 700-1300 °C is more or less the same for all the
three brick qualities. The compression strength increases with temperature up to vicinities of 900-
1000 °C. However, it seems that the compression strength reaches its peak somewhat earlier for
Victor HWM than Silox 60, around 900 °C and 1000 °C respectively. In case of Alex it is less obvious at
which temperature the peak emerges, since only few tests were made, but most probably in the
range of 900-1000 °C.

Silox 60 shows a maximum increase of the crushing strength by more than two times its room
temperature strength, approximately 115 %, while Victor HWM has a lower increase by

34
approximately 65 %. The test of Alex at 1000 °C resulted in an increase of crushing strength by more
than 165 %, but had to be stopped due to precaution reasons.

After the temperature surpasses vicinities of 900 °C for Victor HWM and 1000 °C in the case of Silox
60 and Alex the trend becomes suddenly negative. The compression strength of Silox 60 remains
above the cold crushing test values until the temperature is somewhat over 1100 °C. While for Victor
HWM and Alex the drop below their cold crushing values comes before temperature reaches 1100
°C. Furthermore the graphs reveal that the most rapid strength variation per every 100 °C happens
between 1000 and 1100 °C in cases of all the three bricks. In Chart 4d can be also seen that in the
1000-1300 °C interval the most rapid and significant decrease in compression strength is for Alex.
The strength of Alex at 1300 °C drops by almost 86 % of the room temperature crushing strength or
by around 95 % relatively to its highest registered compression strength at 1000 °C. Victor HWM and
Silox 60 tested at 1300 °C show a decrease of crushing strength by approximately 63 % and 60 %
respectively relative to their cold crushing values or by approximately 78 % and 81 % respectively
relative to their noted maximum crushing strength.

Victor HWM seems to have the slowest decrease of compression strength after it surpasses its peak,
but after the temperature reaches 1200 °C its trend becomes very similar to that of Silox 60. In
general the decreasing trend slows down for all the three brick; for Alex it happens at around 1100
°C, while for Victor HWM and Silox 60 the decreasing trend slows down significantly at around 1200
°C, see Chart 4d. This is also reasonable as the materials become more and more ductile and
approach a more steady state where one mode of deformation (plastic) is dominating (this will be
discussed further down).

35
7.2.1.2 Actual compression strength
Results of high-temperature crushing strength tests of Victor HWM, Silox 60 and Alex are presented
in Chart 5a-c. The charts indicate values obtained at room temperature and in 700-1300 °C interval.
The colored signs denote tests made on the samples from the same batch.

a. b.

c.

Chart 5a-c. Representation of compression strength obtained in RT and 700-1300 °C interval.

One of the important questions to answer is whether the cold crushing test results are relevant to
use in order to relate to a brick’s strength at elevated temperatures. In general it can be said that the
ranking between the high and low- strength bricks at room temperature is also remained at the
higher temperatures. For example, in Chart 5a, regard bricks “17-24v”, “1-5v” and “9-16v” at room
temperature and then at the interval of 1100-1300 °C. It shows clearly that the ranking from the
room temperature remains also at the elevated temperatures. However, it is not fully clear whether
the proportions remain the same or if the differences increase or decrease, but if it changes then it is
probably to a rather low degree. This is also a question about the consistency of the tests, which will
be discussed later. Also, the high increases of compression strength were not expected and in some
cases planned tests in 900-1000 °C range could not be performed due to the limitation of the load
cell and the frame.

By taking into account the general increases from the previous section, it can be estimated that the
maximum strength (at 900-1000 °C) of Victor HWM is above 150 MPa, over 190 MPa for Silox 60 and
over 180 MPa for Alex.

36
7.2.1.3 Soaking of the samples
It is known that the tested materials contain a certain amount of andalusite that is not fully
transformed to mullite. It is also known that fine corundum can react with silica in order to build
secondary mullite. [15]

During the tests with dwell times of 2 h and temperatures maximum 1300 °C no significant reactions
can happen. It requires fairly long time for a significant change to appear at 1300 °C, see Figure 17 for
the example of the kinetics of andalusite transformation. However, it is certain that with time phase
transformations will occur in the new-installed bricks in the rotary-kilns of LKAB.

[22]
Figure 17. Kinetics of mullitization of 200 µm andalusite grains for temperatures ranging from 1200-1600 °C.

One sample of Victor HWM and Silox 60 were placed in 1300 °C during 96 h in order to see if the
treatment would result in any significant change of the crushing strength at 1200 °C. See Chart 6a-b
for the results of the tests.

Chart 6a-b. Hot compression tests of Victor HWM and Silox 60 at 1200 °C made on regular samples and samples that
were additionally soaked at 1300 °C during 96 h. Values in parentheses denote CCS.

Victor HWM shows an increase in strength by approximately 11 % while Silox 60 increases its
strength by over 40 %. The increase for Victor HWM is relatively small and can possibly be within the
natural variation of the material, at least partly. This means of course that the difference could have
also been somewhat bigger. However, it should be remarked that the tested Victor HWM sample is
from a high-strength category, meaning that it is probably already burned to a high degree. The Silox

37
60 sample is from a low/middle- strength category and the strength increase is of a significant
degree. Furthermore, if as-received Silox 60 is prepared under the same conditions as as-received
Victor HWM, then it reasonably contains more unburned andalusite in any case, since andalusite is
the primary base material for Silox 60. It is relevant to expect that the difference in compression
strength after long-time soaking (at sufficiently high temperatures) will be larger for Silox 60 than
Victor HWM in most of the cases.

In Graph 2 XRD-graphs of Silox 60 sample prior and after the soaking in 1300 °C during 96 h can be
found. See Appendix C for corresponding XRD-analysis for Victor HWM sample.

a.

b.

Graph 2. X-ray diffractographs of Silox 60 prior to (a) and after (b) the soaking at 1200 C during 96 h. (Range: 10-90 2θ
degrees. Step width: 0.05 degrees. Time/step: 5 sec.).

The graphs clearly show that the amount of andalusite after the soaking diminishes and that the
intensity of mullite is increased. In both Victor HWM and Silox 60 an unknown peak was detected in
samples prior to the soaking but that disappeared after the soaking. However, this is most probably a
kaolinite peak, but it has not been confirmed.

38
7.2.1.4 Evaluation of the compression strength increase
The origin to the bricks’ dramatic increase of compression strength in the 700-1000 °C interval is
somewhat uncertain. Very little information exists on that subject.

It is reasonable to exclude the influence of phase transformations. Firstly, the materials are already
fired to a high degree, consisting mostly of temperature-stable phases. Secondly, as mentioned
previously, possible phase transformations of significant degree (first of all referring to the
transformation of remained andalusite) require far higher temperatures or/and longer holding times
than those during the tests. However, it is rather certain that the strength decrease after 1000 °C are
due to softening of the vitreous phase. This is also coherent with the results; Alex which contains
highest amount of the glassy phase (19-20 wt%) experiences the largest drop of the strength. While
Victor HWM with smallest amount of the glassy phase (8-10 wt%) have the smallest drop and Silox 60
(14-16 wt%) an intermediate drop in strength. Reasonably, the glassy phase is dominated by
amorphous silica, guideline value of the service temperature of such glass lies in 900-1200 °C interval
(96 % SiO2). [12]

In a way refractory bricks are composite materials, consisting of different phases and grains of
different sizes. The glassy phase functions partly as a binding matrix and has a high influence on the
materials at elevated temperatures. When the bricks are heated the micro constituents will expand.
Considering the three main phases in the bricks; corundum, mullite and the glassy phase it is found
that corundum has the highest thermal expansion coefficient (5-10×10-6/°C), mullite has
intermediate coefficient (3.5-5×10-6/°C) and amorphous silica (assuming is the dominant constituent
in the glassy phase) has the lowest thermal expansion coefficient (< 1×10-6/°C). [12]

It is reasonable to assume that the strength increase is possibly due to the internal stresses in the
material caused by uneven thermal expansion of the constituents- the interlocking effect. Important
information can be found by regarding the differences in strength increase of the three bricks. Why
does Victor HWM increases its strength by approximately 65 % at the most while Silox 60 and Alex by
approximately 115 % and by over 165 % respectively? If the glassy phase functions as a binder within
the materials, then it is obvious that the expansion of corundum and mullite will cause compressive
stresses within the brick, since the glassy phase has much lower thermal expansion coefficient and
will prevent the expansion of the structure. This results in a stronger structure and grain separation
within the bulk material is more difficult to achieve. There is also a coherency between the increase
of the strength and amount of the glassy phase. Victor HWM contains smallest amount of glass and
also has the lowest increase, while Alex has the highest increase in strength and highest amount of
glass. If the theory is true then it is reasonable that a higher amount of glass in the matrix will cause a
stronger structure with higher compressive stresses, when the remaining constituents expand. The
drastic decrease of compression strength after passing 1000 °C is also a sign that confirm that the
glassy phase is the main reason for the structure cohesiveness.

39
7.2.2 Fracture of bricks

7.2.2.1 Deformation modes

The deformation mechanisms at the macro scale are the same for all the three brick types. As the
material compresses, zones with grain separation start to form. Understandably the most affected
zones are radiating from the center of the cylinder to the long side since a lesser amount of the
material have support from surrounding and therefore concentrated tensile stresses can create most
damage here. Typical V-shaped zones can be distinguished in the damaged cylindrical samples, see
Figure 18a. The characteristics of the damage over the interval of tested temperatures are not the
same. At room temperature tests the samples usually did not retain their deformed shape, but were
slowly crushed and material often easily detached. However, at RT-tests samples also occasionally
deformed in an explosive-like way, this was among others noted for the high-strength Victor HWM
(17-24v) and Silox 60 (47-54s), see Figure 18c. The samples at tests in 700-900 °C interval had a big
tendency to explode; this tendency can be related to the proposed theory of the uneven thermal
expansion. The higher compression stresses within the material the higher tendency of sudden
collapse due to external applied loads. No samples at 1000 °C or above had ever exploded.
Additionally, the deformed samples after the tests in 1000-1300 °C keep their deformed shape much
better than the samples tested under 1000 °C (when maximum 10 % strain). This is certainly due to
the softening of the glassy phase that can easier stick together when its viscosity is lowered in
sufficiently high temperatures.

a. b.

c. d.

Figure 18. a.) Cross-section of a damaged sample with a characteristic V-shaped deformation pattern. b.) View from
above of a heavily deformed sample after a RT test. c.) Side view of a remained V-shaped bottom part of a sample after a
RT-test at which the sample exploded. d) Compressed Alex sample at 1300 °C.

By studying strength-displacement graphs it is possible to estimate how bricks behave at different

temperatures. Up to 900 °C the bricks have a typical quasi-brittle fracture mode, without any
significant plastic deformation before collapsing, see Charts 8a-c. This is true for all the three brick

40
qualities. The trends for the materials from the same batch tested up to 900 °C are also very similar,
indicating that stiffness of the bricks is more or less the same up to 900 °C, note the slopes of the
deformation curves in Chart 8a-c. In tests at 1000 °C the materials’ deformation curves start to show
a clear plastic zone. As the temperature increases materials become more and more ductile, see
Charts 8d-f. Deformation graphs reveal that in terms of ductility both Victor HWM and Silox 60
behave in rather similar ways. Also, Silox 60 tends to deform slightly more before it starts to collapse.
However, a big difference in fracture behavior in 1000-1300 °C interval is noted for Alex, which
shows a significant mode transition, losing more or less all of its elasticity already at 1100 °C and the
samples do not collapse as clearly as in case of Victor HWM or Silox 60. Alex seems to behave
viscoplastically in a higher extent than the other brick types. This behavior is certainly primarily due
to higher amount of glassy phase compared to Victor HWM and Silox 60. Furthermore, as mentioned
previously, in case of Silox 60, which is dominated by mullite, large amount of glassy phase is trapped
in capillaries of mullite and therefore the effect of glass softening is less dramatic. This is probably
the reason why Silox 60 behaves similar to Victor HWM even though Silox 60 contains higher amount
of glassy phase.

a. d.

b. e.

c. f.
Chart 8a-e. The tests in a chart are on the samples from the same batch. a-c) Indicate that up to 900 C the bricks fracture
in a typical brittle way. d-f) Show increased ductile behavior of the bricks starting from 1000 °C.

41
7.2.2.2 Crack propagation
In tests up to 1000 °C cracks propagate preferably in the matrix (at macroscopic level) by following
the borders of the large grains, see Figure 19a. This is therefore intergranular crack propagation type.
Where the matrix is consisting of glassy phase and randomly distributed small grains. However also
pure intergranular crack propagation exists (at microscopic level), see Figure 19b. At macroscopic
level these can however be called transgranular cracks, since they go through the large grains and do
not follow the borders of the large grain.

Figure 19. SEM images of crack propagation. a) x30, crack propagation in matrix following grain boundaries (Silox 60
tested in 1000 °C). b) x850, intergranular crack propagation (Victor HWM, RT test).

In the tests above 1000 °C crack propagation is also predominantly in the matrix. One difference is
that the grain separation is more evident in the samples tested below 1000 °C. Not surprisingly
matrix in the materials is rich in silica, due to its high content of glass. The samples that have been
soaked during 96 h in 1300 °C and tested at 1200 °C give a good visualization of how cracks
propagate in the matrix, see Figure 20 and Chart 8.

Mullite-rich in SiO2 2 60

Corundum in TiO2-rich matrix 50

O
40
Atomic %

3 1 SiO2-”island” Al
30
Si
20
Ti
10 Fe
SiO2-rich chanels
0
1 2 3
Spot number

Figure 20 and Chart 8. (left) SEM-image (x850) of micro-crack propagation in the Si-rich matrix in compressed Silox 60
sample at 1200 °C after soaking in 1300 °C during 96 h; surface perpendicular to the applied load. (right) EDS-analysis of
the marked spots in the image to the left.

It can be clearly seen that the glass has been melted and that cracks are radiating in the matrix, while
propagation is blocked by the large grains. Additional SEM-images of micro-crack propagation in Si-
rich matrix with EDS-analysis can be found in Appendix D.

42
7.3 Evaluation of test method and test data
7.3.1 Effect of the powder
The first room temperature tests showed that the measured compression strength was
approximately 15 % below the values obtained by Höganäs Bjuf AB. At that point it was believed to
be either due to natural variation of the bricks strength and/or due to the size difference of the
samples tested by the manufacturer. Many high-temperature tests were already made; therefore it
was decided to keep all the test settings the same for the sake of comparison. However, some
further cold crushing tests showed a large scattering to the values measured by Höganäs Bjuf AB.
This was confusing, but an explanation was found.

As mentioned before alumina powder was placed under the samples in every test with purpose to
distribute the load evenly on the samples in case of small deviations, but mostly in order to avoid the
possible bending stresses on the equipment which is very sensitive to that. In order to see the
influence of alumina powder CCS-tests were made on Silox 60 samples from the same batch. Two
cold crushing tests were performed without using powder. The usually used equipment was removed
and a spherical bearing block was used as a load distributer. Furthermore, two CCS-tests were made
using alumina powder and consequently no spherical bearing block. Additionally, one test without
either spherical bearing block or alumina powder was made. The results are presented in Chart 9.
The black line represents the cold crushing strength obtained by Höganäs Bjuf AB for the tested
batch.

90

80
s47-with powder-no
Compression strength (MPa)

70 bearing

60 s51-with powder-no
Höganäs Bjuf bearing
50 s49-no powder-with
40 bearing
s48-no powder-with
30 bearing
20 s52-no powder-no
bearing
10

Chart 9. Results of CCS-tests made in different compression conditions.

The results are showing a large discrepancy between test methods. Alumina powder is apparently
affecting the true compression strength measurements in a much larger extent than imaginable.
Surprisingly, even measurements by Höganäs Bjuf AB are rather low compared to the maximum
noted strength. One explanation could be that the manufacturer of the bricks makes generally quite
systematical tests and are interested in approximate strength category of the bricks. Therefore
quality of the samples (such as parallelism) is not always according to standard. Additional
explanation can be the size effect; Höganäs Bjuf AB tests cylindrical samples of size 50×50 mm

43
compared to 35×35 mm in these tests. The effect of this has to be investigated, but is probably not
very large.

The effect of powder has to be further studied. The overall conclusion that can be made is that the
powder does not distribute the load evenly on the sample, but it does make sample parallel between
the pushing bars. It is probable that the powder does not give sufficient support close to the border
of the sample. This can be related to the mechanics of Hertzian contact. Which in simple terms
states, for a case similar to the test conditions (“contact between a rigid cylinder and an elastic half-
space”), that pressure under such cylinder is highest close to the border of the sample [23]. There
might be other mechanical reasons too. However, the important discussion is whether the usage of
alumina powder makes tests valid or not.

First of all, the increase in strength due to temperature is certain, large difference in strength
increase between the different brick types is a valid proof. Otherwise the mechanical influence of
powder would have resulted in the same strength increase for every brick type. It is also certain that
the tests are valid for the comparison of the different samples (different batches or types) between
each other tested at the same temperatures, since they do have experienced the same conditions.
However, does alumina powder makes it valid to compare behavior of the samples from the same
batch but tested at different temperatures?

Every test is prepared in the same way; alumina powder has been used under every sample with
approximately the same nominal thickness. The contact area is affected by the powder. But the
strength is calculated on the area of the sample. If the contact area more or less remains during the
whole test procedure then the tests are valid for the comparison. The mechanical conditions have to
be the same in order to compare tests made in different temperatures. However, one fact that has
not been considered is that the powder is influenced by the temperature too, it might start to sinter.
The question is if that influences the contact area. The grain size of used powder is < 0.45 µm and
under the test conditions (time) the considerable sintering of powder is reasonably above 1100 °C or
even more.

Consider the tests in 700-1000 °C interval where the highest increases are noted. The compression
test at 700 °C takes four hours before actual compression, and approximately five hours if tests are
made in 1000 °C. Powder compresses heavily during such long pre-load (1kN) and is affected by the
temperature, it is therefore rigid. It is reasonable to believe that the nominal contact area is “the
same” in this both cases, before the compression takes place. Tests at 900-1000 °C have the highest
increases in strength. But to assume that these increases are due to the larger contact area from the
powder is not reasonable. Mechanical conditions are very equal in both 700 and 1000 °C tests. The
layer of compressed powder is thin, maximum 1mm, and is rigid. It is realistic that the powder effect
is comparable from one temperature to another. Lastly, above 1100 °C the effect of softening of
glassy phase is very large and the compression strength is well reflected.

Tests without powder at elevated temperatures were not performed due to precaution, but also
because the limit of the frame would have probably been reached.

7.3.2 Double peaks

A couple of samples in the 700-900 °C range had a peculiar tendency of having double-peaks; some
of them are presented in Chart 7, where vertical lines denote that the samples exploded. Usually the

44
tests were stopped as soon as the second peak was passed (in order to protect the heating elements
an alumina tube was placed around sample in case of explosion). The double-peaks were noted for
Silox 60 and Victor HWM, but not always and as mentioned only in 700-900 °C range. No such
tendency was noted for Alex, but no certain conclusion about that can be withdrawn for Alex since
only few tests were made for it.

Chart 7. Representing a tendency of “double-peaks” of Victor HWM and Silox 60 samples tested in 700-900 °C.

One possible reason for the double peaks can be the powder below the samples. Assuming that the
powder does not give full support around the whole area of the sample (specifically around the
border of the sample), then the compression force will mostly affect the middle-area of the sample.
But, at some point during the increased load during testing the border of the sample will obtain
sufficient support due to the powder being compressed, therefore the totally applied force can
slightly increase again. This explanation is possible because of the presence of the powder. If both
contact surfaces of the sample were in contact with sufficiently hard load surfaces, that only deform
insignificantly, then this explanation wouldn’t be relevant.

45
7.4 Analysis of contaminated Victor HWM
7.4.1 Analysis of raw materials

7.4.1.1 X-ray diffractometry

Several XRD-scans were made to characterize the crystalline phases in the powders of base
materials. As mentioned previously the base materials were New RASC and Old RASC where the first
mentioned was suspected to be contaminated. The scans of the powders were firstly done in an
aluminum holder, whereupon unknown peaks were found in both raw materials but of significantly
different intensities. This indicated a difference in content of the raw materials. However, these
peaks could not be identified to any relevant candidates. Suspecting a problem with the apparatus
new XRD-scans were made but now in an amorphous holder, by which parasite-peaks could be
avoided. The obtained patterns of the two raw materials were identical and every peak could be
identified. This confirmed that some false peaks were detected when using aluminum holder, which
was due to the un-calibrated X-ray source radiation. Nevertheless, due to this problem, also
significant amorphous peak from the amorphous holder were detected in the range of 5-20 degrees -
hiding possible differences in the raw materials at this range. However, by combining the patterns
obtained from aluminum holder and amorphous holder the false peaks can be identified and the
peaks hiding below the amorphous area can be distinguished. XRD-patterns of New RASC and Old
RASC obtained in aluminum and amorphous holder in the range of 5-90 degrees can be found in
Appendix E. There it can be seen that the whole pattern 5-90 degrees is identical for the two tested
powders. In Graph 3 a XRD-graph representing extra accurate patterns of New RASC and Old RASC in
the range of 20-64 degrees can be found.

46
 Graph 3. XRD-patterns of Old RASC and New RASC. (Range: 20-64 2θ degrees. Step width: 0.05 degrees. Time/step: 5
sec.).

The raw materials showed to be fully calcined and partly mullitized. As can be seen in Graph 3 no
apparent difference between the raw materials can be distinguished at. It was not possible to detect
hematite, rutile or other oxides; only the most common oxides were detected such as corundum,
mullite and andalusite (see Graph 3). The precision of the XRD equipment permits detection of
crystalline phases down to 1-2 wt%. These results show that contaminations are difficult to detect
and confirm by XRD alone.

47
7.4.1.2 Energy-dispersive X-ray spectroscopy
Quantitative EDS-analysis of the powders was made. Six EDS-scans were done on every powder
under identical test set up, covering an area of approximately 0.3×0.3 mm2 per scan. The results of
that are represented in Chart 10a-b.

a. Atomic %. Old RASC - New RASC. b. Weight %. Old RASC - New RASC.
Powders O_OR
40
Powders O_OR
55
O_NR O_NR
50 35
Al_OR Al_OR
45
Al_NR 30 Al_NR
40
Ti_OR Ti_OR
35 25
Ti_NR Ti_NR
30
Si_OR 20 Fe_OR
25
Si_NR Fe_NR
20 15
Fe_OR Si_OR
15 10
Fe_NR Si_NR
10
K_OR 5 K_OR
5
K_NR K_NR
0 0

Chart 10a-b. Summary of the EDS-scans of Old RASC (OR) and New RASC (NR) powders representing atomic (right) and
weight (left) percentages of the six most occurred elements, including standard deviation.

The aluminum content in New RASC showed to be a few percent higher than in Old RASC, while
potassium content was almost twice as high in Old RASC. The rest of the common elements were in
relatively comparable amounts. See Appendix F for the results presented in a table form. The
abundance of other elements which are not presented here can be also found in Appendix D. It is
relevant to state that these elements are of a less importance due to their low concentration.

The higher concentration of aluminum in New RASC relatively to Old RASC can simply mean that the
concentration of aluminum oxide is higher in the first one. A few percent would be of course an
acceptable natural variation. However, when the concentration of a metal increases and the
concentration of oxygen remains the same or gets lower it can indicate that other types of
compounds form in a larger extent. Assuming that most of the metal atoms are bonded to oxygen in
their most preferable state (Al2O3, TiO2, SiO2, Fe2O3, K2O) then a lesser amount of oxygen means that
other compounds have to be formed. This can be shown by a simple proportion calculation, see
Appendix F. In this case it shows that New RASC have a slightly higher (≈ 5%) lack of oxygen, meaning
that a higher amount of other than most preferable oxides have to be formed. However, this is just
an indication and the variation is rather small for the confirmation.

48
7.4.2 Analysis of sintered materials

7.4.2.1 Energy-dispersive X-ray spectroscopy

The EDS-analysis of Victor HWM made from New RASC, contamination-free Victor HWM and EDS-
analysis of the New RASC powder can be found in Chart 11.

60
VHWM
50 VHWM-contaminated
New RASC
40

30

20

10

0
O Al Si Ti Fe K

Chart 11. EDS-analysis of Victor HWM free from contamination; contaminated Victor HWM and New RASC powder.

By comparing EDS-analysis of New RASC and contaminated Victor HWM it can be found that they do
not have the same composition, even though the contaminated Victor HWM is said to be based on
New RASC. The reason for this is that New RASC is the primary raw-materials source, but not the sole
one. In case of Victor HWM the raw material mixture consists approximately of 70 % bauxite, while it
also has added andalusite (≈15%), fine alumina (≈10 %) and clay (≈5 %).

When regarding Victor HWM made from New RASC and ”normal” Victor HWM it can be seen that
the concentration of silica (≈ 40% higher), iron (≈ 45% higher) and potassium (≈ 35% higher) are
higher in the contaminated brick, while the concentration of aluminum (≈ 15% higher) and titanium
(≈ 110% higher) are higher in the ”normal” brick. See Appendix G (Table 1) for the results presented
in a table form. These data and the data from the powder analysis indicate that the difference in
content comes from other sources than the base raw materials (New RASC and Old RASC). E.g. iron
and silica content in “normal” raw material (Old RASC) and “contaminated” raw material (New RASC)
were more or less the same (see Chart 10), but is different in sintered materials (see Chart 11).
Therefore, other sources have to be investigated. That can come from the above mentioned addition
of andalusite, alumina and clay. This can of course also be a question of such things as bad mixing of
the materials.

49
In Figure 21 a picture of a sintered piece of Victor HWM made from New RASC “bauxite” can be
found. Black spots can be visually seen on the surface of it. These are believed to be the
contamination particles. Of course similar abnormalities exist also in “normal” Victor HWM but not in
the same proportions.

Figure 21. Contaminated surface of Victor HWM. (Left) x7. (Right) Zoom in of the contamination particle, magnification
unknown.
Regard SEM-images in Figure 22 that represent the contaminated area. The bright spots are the
contamination particles. The particles are rather similar in size and are approximately 50 µm large.

Further down can be found EDS-analysis of a contamination particle, see Chart 12. The sequences of
the scanning can be seen in Figure 22, right picture, marked with black crosses; count from right to
left, each cross corresponds to an EDS-scan.

Figure 22. (Left) SEM image of the contaminated area in Victor HWM, x45. (Right) Magnification of the marked zone,
x180. BSE-mode.

50
 EDS-analysis. Contamination particle (Fe-rich)
70
Fe
60
O
50
Al
Atomic % 40 Si
30 Ti

20 Cr

10 K

0 Ca
1 2 3 4 5 6 7 8
Spot number

Chart 12. Sequence of EDS-analysis of a contamination particle rich in iron found in Victor HWM (manufactured from
New RASC), see Figure 22 for reference.

As Figure 22 (right) reveals the first scan is made in the centrum of the contamination particle and
the rest are coherently carried out further and further away from the center. Evidently the
contamination particle is rich in iron. As the distance from the particle’s center increases the iron
concentration drops, while concentration of oxygen, aluminum and silicon increases. Also chromium,
which is a natural follower of iron, is present in and around the particle with rather constant
concentration of around 1.5 %. Also amount of titanium is rather constant through the EDS-
sequence, around 2.5-3 %. Presence of potassium is constant and quite low, while concentration of
calcium is slowly increasing with distance and reaches approximately 1 % in the final scan. See
Appendix G (Table 2) for the results presented in a table form.

Also another type of possible contamination particle was existing in Victor HWM made from New
RASC, see Figure 23a. These were not as common as the first mentioned contamination particles.
Chart 13 represents the EDS-analysis of the particle, while SEM-image of the scanned areas can be
found in Figure 23 b.

Spectrum 2
Spectrum 3

Spectrum 1

b.
a.

Figure 23. a) Contamination particle in Victor HWM. x45. b) SEM image of the EDS-scanned areas of the contamination
particle, x400.

Three scans are made starting from near the centrum. The black area on the right-hand side of the
image displays the corner of the “crater” that can be seen in Figure 23a.

51
 EDS-analysis. Contamination particle (Si-rich)
70

60 O
50 Al
Si
Atomic %
40
Fe
30 K
20 Ca
Na
10
Ti
0
1 2 3
Spot number

Chart 13. Sequence of EDS-analysis of a contamination particle rich in silica found in Victor HWM (manufactured from
New RASC), see Figure 23b for reference.

This type of contamination showed to be primarily rich in silica and potassium, but relatively low in
aluminum. This is probably a particle rich in glass. See Appendix G (Table 3) for the results presented
in a table form.

52
8 Summary and conclusions
High-temperature compression test is a rare characterization method. It describes the strength of the
material in its service temperatures. The performed tests have showed that two different brick types
with the same nominal room temperature strength can have completely different strength at higher
temperatures; therefore high temperature strength is highly ruled by material composition.

In this work has been shown that for the tested materials (Victor HWM, Silox 60 and Alex):
• The compression strength increases with temperature up to approximately 900-1000 °C
• The highest increases are over 60 % for Victor HWM, over 115 % for Silox 60 and over 160 %
for Alex at their strength peaks (900-1000 °C)
• Above 1000 °C the compression strength starts to decrease
• At 1300 °C the compression strength is decreased by approximately 60% for Victor HWM and
Silox 60; and over 80 % for Alex (relative to their RT-compression strength)
• Victor HWM has the smallest compression strength variation over the tested temperature
interval, while Alex has the highest variation
• Compression strength ranking (for the same brick type) at RT remains at elevated
temperatures

Strength behavior at different temperatures is highly affected by the glassy phase in the materials. It
has been seen that:
• Quasi-brittle mode is active until around 1000 °C
• After 1100 °C plastic deformation is dominating
• Crack propagation is predominantly of intergraluar type, via the matrix, in the whole tested
temperature range (RT, 700-1300 °C)
• Crack propagation seems to be in the Si-rich regions mainly

It has also been shown that long-time soaking of the as-received bricks can significantly increase (>40
%) the strength of the bricks. This is primarily due to transformation of remained andalusite to
mullite.

The analysis of raw materials (new RASC and Old RASC) by XRD could not indicate any certain
difference. EDS-analysis of raw materials detected slightly higher amount of aluminum in New RASC
and almost twice as high potassium content in Old RASC compared to New RASC. The rest of the
common elements were in relatively comparable amounts.

Contaminated Victor HWM compared to “normal” Victor HWM showed to have higher amount of
silicon (≈40 % higher), iron (≈45 % higher) and potassium (≈35 % higher). Aluminum content in
“normal” Victor HWM was slightly higher (≈15 % higher), while titanium content was significantly
higher (≈110 % higher), compared to contaminated Victor HWM. These data and the data from the
powder analysis indicate that the difference in content comes from other sources than the base raw
materials (New RASC and Old RASC). E.g. iron and silicon content in “normal” raw material (Old RASC)
and “contaminated” raw material (New RASC) were more or less the same, but is different in
sintered materials. Therefore, other sources can be the reason to contamination.

EDS-analysis of possible contamination particles indicates that two types are common in the
contaminated Victor HWM: iron-rich and silica-rich particles.

53
References
[1+. W.D.Callister, “Materials Science and Engineering an Introduction”, 7th ed., 2007.

[2]. S.C. Carniglia, G.L. Barna, “Handbook of Industrial Refractories. Principles, Types, Properties and
Apllications”, Ch.1, p. 1-24, 1992.

[3]. Harbison-Walker Refractories Division of INDRESCO Inc.. Handbook of refractory Practice. First
edition, 1992.

[4]. S.C. Carniglia, G.L. Barna, “Handbook of Industrial Refractories. Principles, Types, Properties and
Apllications”, Ch.13, p. 457-468, 1992.

[5]. Annual Book of ASTM Standrads, “Standard Classification of Silica Refractory Brick”, C416-97,
Vol.15.01, p. 80, 1998.

[6]. F. Cardarelli, “Materials Handbook. A concise desktop reference”, 2nd ed., 2008.

[7]. Annual Book of ASTM Standrads, “Standard Classification of Fireclay and High-Alumina Refractory
Brick”, C27- 98, Vol.15.01, p. 14-15, 1998.

[8]. Annual Book of ASTM Standrads, “Standard Classification of Insulating Firebrick”, C155-97,
Vol.15.01, p. 43, 1998.

[9]. P.C. Dokko, J.A. Pask, K.S. Mazdiyasni, “High-Temperature Mechanical Properties of Mullite under
Compression”, Journal of the American Ceramic Society, Vol.60, No.3-4, 1977.

*10+. H. Schneider, J. Schreuer, B. Hildmann, “Structure and Properties of Mullite- A Review”, Journal
of the European Ceramic Society, Vol. 28, p. 329-344, 2008.

[11]. http://www.britannica.com/EBchecked/media/2938/Phase-diagram-of-the-alumina-silica-
system, “Aluminosilicate: Phase Diagram of the Alumina-Silica System”, Encyclopaedia Britannica
Online, 2011.

[12]. CES EduPack 2005.

[13]. A. Putnis, “Introduction to Mineral Science”, Ch.6, p.147-150, 1992.

[14]. H. Schneider, K. Okada, J.A. Pask, ”Mullite and Mullite Ceramics”, 1994.

[15]. R.Hagen, C.Lamy, B.Myhre, “Mulite Formation, High-Temperature Properties of Andalusite-

Alumina Based LCC and ULCC”, Elkem Materials, 2009.

[16]. Annual Book of ASTM Standrads, “Standard Test Method for Cold Crushing Strength and
Modulus of Rapture of Refractories”, C133-97, Vol.15.01, p. 31-33, 1998.

*17+. M.A. van Gils, “Quasi-Brittle Fracture of Ceramics”, 1997.

*18+. C.Li, E.Nordlund, “ Deformation of Brittle Rocks Under Compression-with Particular Reference
to Microcracks”, Mechanics of Materials, 15, p. 223-239, 1993.

54
*19+. E.Z.Wang, N.G. Shrive, “Brittle Fracture in Compression: Mechanics, Models and Criteria”,
Engineering Fracture Mechanics, Vol.52, No.6, p. 1107-1126, 1995.

[20]. E.Z. Lajtai, B.J. Carter, M.L Ayari, “Criteria for Brittle Fracture in Compression. Engineering
Fracture Mechanics”, Vol. 37, No.1, p. 59-74, 1990.

*21+. J. Stjernberg, “Degradation of mullite based materials by alkali containing slags”, Licentiate
Thesis, Luleå University of Technology, 2008.

[22]. M-L. Bouchetou, J-P. Ildefonse, J. Poirier, “Mullite grown from fired andalusite grains: the role of
impurities and of the high temperature liquid phase on the kinetics of mullitization and
consequences on thermal shocks resistance”, Ceramics International, Vol. 31, p. 999-1005, 2005.

[23]. I. N. Sneddon, “The Relation between Load and Penetration in the Axisymmetric Boussinesq
Problem for a Punch of Arbitrary Profile”,Int. J. Eng. Sci. v. 3, pp. 47–57, 1965.

55
Appendix A

3 2
1

a) b)

Figure 1. a) Picture of the compression equipment set up. Notations: 1: Al2O3-support; 2: sample; 3: Al2O3-protection ring.
b) Image of the misalignment of the alumina pushing bar.

b)

Figure 2. a) Water distributor. b) Steel sheet with an on-

glued isolation layer.
a)

i
Appendix B
Bending strength calculations by “piston-on-three-ball”-test.

( )
(1)

( ) (2)
( ) ( ) ( )

(3)
( )( ( ) ) ( )( )

where

σ: maximum center tensile stress. Flexural strength (MPa). (500)

P: load at fracture (N). (100000)
b: thickness of the specimen (mm). (to find)
ν: Poisson’s ratio. (0.25)
r1: radius of support circle (mm). (22.5)
r2: radius of loaded area (radius of the compression head) (mm). (17.5)
r3: radius of the specimen (mm). (23)

From (1) can be found the minimum thickness required for the ceramic support (see Figure 1,
Appendix A). In this case it was found to be approximately 11 mm. The calculations are for the
exactly the same conditions as during the performed tests but can be used in order to define a lower
limit of the thickness.

ii
Appendix C

a.

b.

Graph 1. X-ray diffractographs on the Victor HWM sample prior (a) and after (b) the soaking at 1200 °C during 96 h.

iii
Appendix D

60
1
50 O
3
40 Al

30 Si
2
Ti
20
4
Fe
10
K
0
c. 1 2 3 4
a.

4
70
60
O
50
Al
3 40
Si
30
2 Ti
20
Fe
1 10
K
0
d. 1 2 3 4
b.

Figure 1a-b. Chart 1c-d. a (x250) and b (x500): SEM-images of crack propagation in Si-rich regions in Victor HWM after
compression at 1200 °C. c and d: EDS-analysis of the marked zones.

iv
Appendix E

Graph 1. X-ray diffractograms of old RASC and New RASC measured in different holders. (Range: 20-64 2θ degrees. Step
width: 0.05 degrees. Time/step: 5 sec.)

v
 Appendix F

Atomic % Weight %
Old RASC New RASC Old RASC New RASC
Element Average Standard Average Standard Average Standard Average Standard
deviation deviation deviation deviation
O 53,44 0,818 51,48 0,948 35,34 0,874 33,91 0,942
Al 28,18 0,099 31,56 0,564 31,42 0,400 35,06 0,771
Si 5,13 0,495 4,37 0,613 5,95 0,581 5,052 0,681
Ti 7,78 0,531 7,53 0,434 15,40 0,967 14,84 0,787
Fe 4,01 0,538 4,19 0,323 9,26 1,150 9,62 0,669
K 0,885 0,174 0,49 0,180 1,43 0,284 0,79 0,281

Table 1. Results of EDS-analysis representing the average concentration of the elements present in Old RASC and New
RASC powders.

Atomic %
Old RASC New RASC
Element spot 6 spot 5 spot 4 spot 3 spot 2 spot 1 spot 6 spot 5 spot 4 spot 3 spot 2 spot 1
V 0,41 0,3 0,23 0,34 0,24 0,52
Cr 0,37 0,22 0,23 0,36 0,31 0,38 0,2
Ni 0,66
Ca 0,21 0,18
Na 0,36

Table 2. Results of EDS-analysis representing the abundance of the minor elements present in Old RASC and New RASC
powders.

Proportion calculation
Preferable oxide state of a Number of oxygen atoms per Required number of oxygen Required number of oxygen
metal metal atom atoms. Old RASC atoms. New RASC
Al2O3 3/2 28,18 × 3/2 31,56 × 3/2
SiO2 2 5,13 × 2 4,37 × 2
TiO2 2 7,78 × 2 7,53 × 2
Fe2O3 3/2 4,01 × 3/2 4,19 × 3/2
K2O 1/2 0,885 × 1/2 0,49 × 1/2
Total 74,55 77,67

Values are collected form Table 1.

Old RASC: (74,55-53,44)/74,55=28,3 %

New RASC: (77,67-51,48)/77,67=33,7 %

This means that 28,3 % and 33,7 % of the required oxygen in Old RASC and New RASC respectively is
actually not available in order to build the preferred metal oxides only. Consequently it means that
other compounds than preferred metal oxides are existing in the powders and in this case in a
slightly larger extent in New RASC than in Old RASC.

vi
Appendix G

Atomic %
Victor HWM Victor HWM-contaminated
Element Average Standard deviation Average Standard deviation
O 47,48 0,63 47,134 1,604967
Al 31,495 0,045 27,53 0,805854
Si 8,94 0,49 12,684 0,77999
Ti 4,945 0,165 2,352 0,274255
Fe 4,575 0,025 6,646 1,211917
K 0,95 0,07 1,314 0,14207

Table 1. EDS-analysis made on “normal” Victor HWM and Victor HWM manufactured from New RASC that is
contaminated.

Atomic %
Element Spot 1 Spot 2 Spot 3 Spot 4 Spot 5 Spot 6 Spot 7 Spot 8
O 16,98 25,28 20,08 27,54 40,5 43,04 39,27 43,59
Al 16,47 25,94 27,91 36,18 30,3 30,4 31,01 30,62
Si 1,87 2,78 2,92 6,34 6,88 10,29 11,02 8,92
Ti 3,45 2,46 2,29 2,04 4,47 2,47 3,13 2,48
Fe 58,97 41,81 44,18 23,26 15,39 11,42 11,99 10,99
K 0,16 0,14 0,26 0,48 0,21 0,24 0,08 0,64
Ca 0,37 0,07 0,24 0,43 0,59 0,67 0,87 0,92
Cr 1,07 1,43 1,59 3,28 1,35 1,34 1,65 1,73
Ni 0,19 0,08 0,16 0,18 0,12 0,02 0,09

Table 2. Result of a sequence of EDS-analysis made on a suspected contamination particle from Victor HWM
manufactured from New RASC-“bauxite”. See Figure 14 (right) for reference.

Atomic %
Element Spot 1 Spot 2 Spot 3
O 48,47 60,87 49,96
Al 9,61 12,31 28,37
Si 26,66 18,99 13,58
Ti 0,61 1,78
Fe 4,53 2,17 3,72
K 5,9 2,94 1,23
Ca 2,41 1,26 0,65
Cr 0,44
Na 1,86 1,03 0,28

Table 3. EDS-analysis made on a suspected contamination particle from Victor HWM manufactured from New RASC. See
Figure 16 for reference.

vii